bus.c 30.1 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13
/*
 * Copyright(c) 2013-2015 Intel Corporation. All rights reserved.
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of version 2 of the GNU General Public License as
 * published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful, but
 * WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License for more details.
 */
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
14
#include <linux/libnvdimm.h>
15
#include <linux/sched/mm.h>
16
#include <linux/vmalloc.h>
17
#include <linux/uaccess.h>
18
#include <linux/module.h>
19
#include <linux/blkdev.h>
20
#include <linux/fcntl.h>
21
#include <linux/async.h>
22
#include <linux/genhd.h>
23
#include <linux/ndctl.h>
24
#include <linux/sched.h>
25 26 27
#include <linux/slab.h>
#include <linux/fs.h>
#include <linux/io.h>
28
#include <linux/mm.h>
29
#include <linux/nd.h>
30
#include "nd-core.h"
31
#include "nd.h"
32
#include "pfn.h"
33

34
int nvdimm_major;
35 36
static int nvdimm_bus_major;
static struct class *nd_class;
37
static DEFINE_IDA(nd_ida);
38

39 40 41 42
static int to_nd_device_type(struct device *dev)
{
	if (is_nvdimm(dev))
		return ND_DEVICE_DIMM;
43
	else if (is_memory(dev))
44 45 46
		return ND_DEVICE_REGION_PMEM;
	else if (is_nd_blk(dev))
		return ND_DEVICE_REGION_BLK;
47 48
	else if (is_nd_dax(dev))
		return ND_DEVICE_DAX_PMEM;
49
	else if (is_nd_region(dev->parent))
50
		return nd_region_to_nstype(to_nd_region(dev->parent));
51 52 53 54 55 56

	return 0;
}

static int nvdimm_bus_uevent(struct device *dev, struct kobj_uevent_env *env)
{
57 58 59 60
	/*
	 * Ensure that region devices always have their numa node set as
	 * early as possible.
	 */
61
	if (is_nd_region(dev))
62
		set_dev_node(dev, to_nd_region(dev)->numa_node);
63 64 65 66
	return add_uevent_var(env, "MODALIAS=" ND_DEVICE_MODALIAS_FMT,
			to_nd_device_type(dev));
}

67 68 69
static struct module *to_bus_provider(struct device *dev)
{
	/* pin bus providers while regions are enabled */
70
	if (is_nd_region(dev)) {
71 72
		struct nvdimm_bus *nvdimm_bus = walk_to_nvdimm_bus(dev);

73
		return nvdimm_bus->nd_desc->module;
74 75 76 77
	}
	return NULL;
}

78 79 80 81 82 83 84 85 86 87 88
static void nvdimm_bus_probe_start(struct nvdimm_bus *nvdimm_bus)
{
	nvdimm_bus_lock(&nvdimm_bus->dev);
	nvdimm_bus->probe_active++;
	nvdimm_bus_unlock(&nvdimm_bus->dev);
}

static void nvdimm_bus_probe_end(struct nvdimm_bus *nvdimm_bus)
{
	nvdimm_bus_lock(&nvdimm_bus->dev);
	if (--nvdimm_bus->probe_active == 0)
89
		wake_up(&nvdimm_bus->wait);
90 91 92
	nvdimm_bus_unlock(&nvdimm_bus->dev);
}

93 94 95
static int nvdimm_bus_probe(struct device *dev)
{
	struct nd_device_driver *nd_drv = to_nd_device_driver(dev->driver);
96
	struct module *provider = to_bus_provider(dev);
97 98 99
	struct nvdimm_bus *nvdimm_bus = walk_to_nvdimm_bus(dev);
	int rc;

100 101 102
	if (!try_module_get(provider))
		return -ENXIO;

D
Dan Williams 已提交
103 104 105
	dev_dbg(&nvdimm_bus->dev, "START: %s.probe(%s)\n",
			dev->driver->name, dev_name(dev));

106
	nvdimm_bus_probe_start(nvdimm_bus);
107
	rc = nd_drv->probe(dev);
108 109
	if (rc == 0)
		nd_region_probe_success(nvdimm_bus, dev);
110 111
	else
		nd_region_disable(nvdimm_bus, dev);
112 113
	nvdimm_bus_probe_end(nvdimm_bus);

D
Dan Williams 已提交
114
	dev_dbg(&nvdimm_bus->dev, "END: %s.probe(%s) = %d\n", dev->driver->name,
115
			dev_name(dev), rc);
116

117 118
	if (rc != 0)
		module_put(provider);
119 120 121 122 123 124
	return rc;
}

static int nvdimm_bus_remove(struct device *dev)
{
	struct nd_device_driver *nd_drv = to_nd_device_driver(dev->driver);
125
	struct module *provider = to_bus_provider(dev);
126
	struct nvdimm_bus *nvdimm_bus = walk_to_nvdimm_bus(dev);
127
	int rc = 0;
128

129 130
	if (nd_drv->remove)
		rc = nd_drv->remove(dev);
131 132
	nd_region_disable(nvdimm_bus, dev);

133 134
	dev_dbg(&nvdimm_bus->dev, "%s.remove(%s) = %d\n", dev->driver->name,
			dev_name(dev), rc);
135
	module_put(provider);
136 137 138
	return rc;
}

139 140 141 142 143 144 145 146 147 148 149 150 151 152 153
static void nvdimm_bus_shutdown(struct device *dev)
{
	struct nvdimm_bus *nvdimm_bus = walk_to_nvdimm_bus(dev);
	struct nd_device_driver *nd_drv = NULL;

	if (dev->driver)
		nd_drv = to_nd_device_driver(dev->driver);

	if (nd_drv && nd_drv->shutdown) {
		nd_drv->shutdown(dev);
		dev_dbg(&nvdimm_bus->dev, "%s.shutdown(%s)\n",
				dev->driver->name, dev_name(dev));
	}
}

154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179
void nd_device_notify(struct device *dev, enum nvdimm_event event)
{
	device_lock(dev);
	if (dev->driver) {
		struct nd_device_driver *nd_drv;

		nd_drv = to_nd_device_driver(dev->driver);
		if (nd_drv->notify)
			nd_drv->notify(dev, event);
	}
	device_unlock(dev);
}
EXPORT_SYMBOL(nd_device_notify);

void nvdimm_region_notify(struct nd_region *nd_region, enum nvdimm_event event)
{
	struct nvdimm_bus *nvdimm_bus = walk_to_nvdimm_bus(&nd_region->dev);

	if (!nvdimm_bus)
		return;

	/* caller is responsible for holding a reference on the device */
	nd_device_notify(&nd_region->dev, event);
}
EXPORT_SYMBOL_GPL(nvdimm_region_notify);

180 181 182 183 184 185 186 187 188 189 190 191
struct clear_badblocks_context {
	resource_size_t phys, cleared;
};

static int nvdimm_clear_badblocks_region(struct device *dev, void *data)
{
	struct clear_badblocks_context *ctx = data;
	struct nd_region *nd_region;
	resource_size_t ndr_end;
	sector_t sector;

	/* make sure device is a region */
192
	if (!is_memory(dev))
193 194 195 196 197 198 199 200 201 202 203 204 205
		return 0;

	nd_region = to_nd_region(dev);
	ndr_end = nd_region->ndr_start + nd_region->ndr_size - 1;

	/* make sure we are in the region */
	if (ctx->phys < nd_region->ndr_start
			|| (ctx->phys + ctx->cleared) > ndr_end)
		return 0;

	sector = (ctx->phys - nd_region->ndr_start) / 512;
	badblocks_clear(&nd_region->bb, sector, ctx->cleared / 512);

206 207 208
	if (nd_region->bb_state)
		sysfs_notify_dirent(nd_region->bb_state);

209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227
	return 0;
}

static void nvdimm_clear_badblocks_regions(struct nvdimm_bus *nvdimm_bus,
		phys_addr_t phys, u64 cleared)
{
	struct clear_badblocks_context ctx = {
		.phys = phys,
		.cleared = cleared,
	};

	device_for_each_child(&nvdimm_bus->dev, &ctx,
			nvdimm_clear_badblocks_region);
}

static void nvdimm_account_cleared_poison(struct nvdimm_bus *nvdimm_bus,
		phys_addr_t phys, u64 cleared)
{
	if (cleared > 0)
228
		badrange_forget(&nvdimm_bus->badrange, phys, cleared);
229 230 231 232 233

	if (cleared > 0 && cleared / 512)
		nvdimm_clear_badblocks_regions(nvdimm_bus, phys, cleared);
}

234 235 236 237 238 239 240 241
long nvdimm_clear_poison(struct device *dev, phys_addr_t phys,
		unsigned int len)
{
	struct nvdimm_bus *nvdimm_bus = walk_to_nvdimm_bus(dev);
	struct nvdimm_bus_descriptor *nd_desc;
	struct nd_cmd_clear_error clear_err;
	struct nd_cmd_ars_cap ars_cap;
	u32 clear_err_unit, mask;
242
	unsigned int noio_flag;
243 244 245 246 247 248
	int cmd_rc, rc;

	if (!nvdimm_bus)
		return -ENXIO;

	nd_desc = nvdimm_bus->nd_desc;
249 250 251 252
	/*
	 * if ndctl does not exist, it's PMEM_LEGACY and
	 * we want to just pretend everything is handled.
	 */
253
	if (!nd_desc->ndctl)
254
		return len;
255 256 257 258

	memset(&ars_cap, 0, sizeof(ars_cap));
	ars_cap.address = phys;
	ars_cap.length = len;
259
	noio_flag = memalloc_noio_save();
260 261
	rc = nd_desc->ndctl(nd_desc, NULL, ND_CMD_ARS_CAP, &ars_cap,
			sizeof(ars_cap), &cmd_rc);
262
	memalloc_noio_restore(noio_flag);
263 264 265 266 267 268 269 270 271 272 273 274 275 276
	if (rc < 0)
		return rc;
	if (cmd_rc < 0)
		return cmd_rc;
	clear_err_unit = ars_cap.clear_err_unit;
	if (!clear_err_unit || !is_power_of_2(clear_err_unit))
		return -ENXIO;

	mask = clear_err_unit - 1;
	if ((phys | len) & mask)
		return -ENXIO;
	memset(&clear_err, 0, sizeof(clear_err));
	clear_err.address = phys;
	clear_err.length = len;
277
	noio_flag = memalloc_noio_save();
278 279
	rc = nd_desc->ndctl(nd_desc, NULL, ND_CMD_CLEAR_ERROR, &clear_err,
			sizeof(clear_err), &cmd_rc);
280
	memalloc_noio_restore(noio_flag);
281 282 283 284
	if (rc < 0)
		return rc;
	if (cmd_rc < 0)
		return cmd_rc;
285

286
	nvdimm_account_cleared_poison(nvdimm_bus, phys, clear_err.cleared);
287

288 289 290 291
	return clear_err.cleared;
}
EXPORT_SYMBOL_GPL(nvdimm_clear_poison);

292 293
static int nvdimm_bus_match(struct device *dev, struct device_driver *drv);

294
static struct bus_type nvdimm_bus_type = {
295
	.name = "nd",
296 297 298 299
	.uevent = nvdimm_bus_uevent,
	.match = nvdimm_bus_match,
	.probe = nvdimm_bus_probe,
	.remove = nvdimm_bus_remove,
300
	.shutdown = nvdimm_bus_shutdown,
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
static void nvdimm_bus_release(struct device *dev)
{
	struct nvdimm_bus *nvdimm_bus;

	nvdimm_bus = container_of(dev, struct nvdimm_bus, dev);
	ida_simple_remove(&nd_ida, nvdimm_bus->id);
	kfree(nvdimm_bus);
}

static bool is_nvdimm_bus(struct device *dev)
{
	return dev->release == nvdimm_bus_release;
}

struct nvdimm_bus *walk_to_nvdimm_bus(struct device *nd_dev)
{
	struct device *dev;

	for (dev = nd_dev; dev; dev = dev->parent)
		if (is_nvdimm_bus(dev))
			break;
	dev_WARN_ONCE(nd_dev, !dev, "invalid dev, not on nd bus\n");
	if (dev)
		return to_nvdimm_bus(dev);
	return NULL;
}

struct nvdimm_bus *to_nvdimm_bus(struct device *dev)
{
	struct nvdimm_bus *nvdimm_bus;

	nvdimm_bus = container_of(dev, struct nvdimm_bus, dev);
	WARN_ON(!is_nvdimm_bus(dev));
	return nvdimm_bus;
}
EXPORT_SYMBOL_GPL(to_nvdimm_bus);

340 341 342 343 344 345
struct nvdimm_bus *nvdimm_to_bus(struct nvdimm *nvdimm)
{
	return to_nvdimm_bus(nvdimm->dev.parent);
}
EXPORT_SYMBOL_GPL(nvdimm_to_bus);

346 347 348 349 350 351 352 353 354 355 356
struct nvdimm_bus *nvdimm_bus_register(struct device *parent,
		struct nvdimm_bus_descriptor *nd_desc)
{
	struct nvdimm_bus *nvdimm_bus;
	int rc;

	nvdimm_bus = kzalloc(sizeof(*nvdimm_bus), GFP_KERNEL);
	if (!nvdimm_bus)
		return NULL;
	INIT_LIST_HEAD(&nvdimm_bus->list);
	INIT_LIST_HEAD(&nvdimm_bus->mapping_list);
357
	init_waitqueue_head(&nvdimm_bus->wait);
358 359
	nvdimm_bus->id = ida_simple_get(&nd_ida, 0, 0, GFP_KERNEL);
	mutex_init(&nvdimm_bus->reconfig_mutex);
360
	badrange_init(&nvdimm_bus->badrange);
361 362 363 364 365 366 367 368 369
	if (nvdimm_bus->id < 0) {
		kfree(nvdimm_bus);
		return NULL;
	}
	nvdimm_bus->nd_desc = nd_desc;
	nvdimm_bus->dev.parent = parent;
	nvdimm_bus->dev.release = nvdimm_bus_release;
	nvdimm_bus->dev.groups = nd_desc->attr_groups;
	nvdimm_bus->dev.bus = &nvdimm_bus_type;
370
	nvdimm_bus->dev.of_node = nd_desc->of_node;
371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401
	dev_set_name(&nvdimm_bus->dev, "ndbus%d", nvdimm_bus->id);
	rc = device_register(&nvdimm_bus->dev);
	if (rc) {
		dev_dbg(&nvdimm_bus->dev, "registration failed: %d\n", rc);
		goto err;
	}

	return nvdimm_bus;
 err:
	put_device(&nvdimm_bus->dev);
	return NULL;
}
EXPORT_SYMBOL_GPL(nvdimm_bus_register);

void nvdimm_bus_unregister(struct nvdimm_bus *nvdimm_bus)
{
	if (!nvdimm_bus)
		return;
	device_unregister(&nvdimm_bus->dev);
}
EXPORT_SYMBOL_GPL(nvdimm_bus_unregister);

static int child_unregister(struct device *dev, void *data)
{
	/*
	 * the singular ndctl class device per bus needs to be
	 * "device_destroy"ed, so skip it here
	 *
	 * i.e. remove classless children
	 */
	if (dev->class)
402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419
		return 0;

	if (is_nvdimm(dev)) {
		struct nvdimm *nvdimm = to_nvdimm(dev);
		bool dev_put = false;

		/* We are shutting down. Make state frozen artificially. */
		nvdimm_bus_lock(dev);
		nvdimm->sec.state = NVDIMM_SECURITY_FROZEN;
		if (test_and_clear_bit(NDD_WORK_PENDING, &nvdimm->flags))
			dev_put = true;
		nvdimm_bus_unlock(dev);
		cancel_delayed_work_sync(&nvdimm->dwork);
		if (dev_put)
			put_device(dev);
	}
	nd_device_unregister(dev, ND_SYNC);

420 421 422
	return 0;
}

423
static void free_badrange_list(struct list_head *badrange_list)
424
{
425
	struct badrange_entry *bre, *next;
426

427 428 429
	list_for_each_entry_safe(bre, next, badrange_list, list) {
		list_del(&bre->list);
		kfree(bre);
430
	}
431
	list_del_init(badrange_list);
432 433 434 435 436 437 438 439 440 441
}

static int nd_bus_remove(struct device *dev)
{
	struct nvdimm_bus *nvdimm_bus = to_nvdimm_bus(dev);

	mutex_lock(&nvdimm_bus_list_mutex);
	list_del_init(&nvdimm_bus->list);
	mutex_unlock(&nvdimm_bus_list_mutex);

442 443 444
	wait_event(nvdimm_bus->wait,
			atomic_read(&nvdimm_bus->ioctl_active) == 0);

445 446 447
	nd_synchronize();
	device_for_each_child(&nvdimm_bus->dev, NULL, child_unregister);

448 449 450
	spin_lock(&nvdimm_bus->badrange.lock);
	free_badrange_list(&nvdimm_bus->badrange.list);
	spin_unlock(&nvdimm_bus->badrange.lock);
451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485

	nvdimm_bus_destroy_ndctl(nvdimm_bus);

	return 0;
}

static int nd_bus_probe(struct device *dev)
{
	struct nvdimm_bus *nvdimm_bus = to_nvdimm_bus(dev);
	int rc;

	rc = nvdimm_bus_create_ndctl(nvdimm_bus);
	if (rc)
		return rc;

	mutex_lock(&nvdimm_bus_list_mutex);
	list_add_tail(&nvdimm_bus->list, &nvdimm_bus_list);
	mutex_unlock(&nvdimm_bus_list_mutex);

	/* enable bus provider attributes to look up their local context */
	dev_set_drvdata(dev, nvdimm_bus->nd_desc);

	return 0;
}

static struct nd_device_driver nd_bus_driver = {
	.probe = nd_bus_probe,
	.remove = nd_bus_remove,
	.drv = {
		.name = "nd_bus",
		.suppress_bind_attrs = true,
		.bus = &nvdimm_bus_type,
		.owner = THIS_MODULE,
		.mod_name = KBUILD_MODNAME,
	},
486 487
};

488 489 490 491 492 493 494 495 496 497
static int nvdimm_bus_match(struct device *dev, struct device_driver *drv)
{
	struct nd_device_driver *nd_drv = to_nd_device_driver(drv);

	if (is_nvdimm_bus(dev) && nd_drv == &nd_bus_driver)
		return true;

	return !!test_bit(to_nd_device_type(dev), &nd_drv->type);
}

498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514
static ASYNC_DOMAIN_EXCLUSIVE(nd_async_domain);

void nd_synchronize(void)
{
	async_synchronize_full_domain(&nd_async_domain);
}
EXPORT_SYMBOL_GPL(nd_synchronize);

static void nd_async_device_register(void *d, async_cookie_t cookie)
{
	struct device *dev = d;

	if (device_add(dev) != 0) {
		dev_err(dev, "%s: failed\n", __func__);
		put_device(dev);
	}
	put_device(dev);
515 516
	if (dev->parent)
		put_device(dev->parent);
517 518 519 520 521 522
}

static void nd_async_device_unregister(void *d, async_cookie_t cookie)
{
	struct device *dev = d;

D
Dan Williams 已提交
523 524 525 526
	/* flush bus operations before delete */
	nvdimm_bus_lock(dev);
	nvdimm_bus_unlock(dev);

527 528 529 530
	device_unregister(dev);
	put_device(dev);
}

531
void __nd_device_register(struct device *dev)
532
{
533 534
	if (!dev)
		return;
535
	dev->bus = &nvdimm_bus_type;
536 537
	if (dev->parent)
		get_device(dev->parent);
538 539 540 541
	get_device(dev);
	async_schedule_domain(nd_async_device_register, dev,
			&nd_async_domain);
}
542 543 544 545 546 547

void nd_device_register(struct device *dev)
{
	device_initialize(dev);
	__nd_device_register(dev);
}
548 549 550 551
EXPORT_SYMBOL(nd_device_register);

void nd_device_unregister(struct device *dev, enum nd_async_mode mode)
{
552 553
	bool killed;

554 555
	switch (mode) {
	case ND_ASYNC:
556 557 558 559 560 561 562 563 564
		/*
		 * In the async case this is being triggered with the
		 * device lock held and the unregistration work needs to
		 * be moved out of line iff this is thread has won the
		 * race to schedule the deletion.
		 */
		if (!kill_device(dev))
			return;

565 566 567 568 569
		get_device(dev);
		async_schedule_domain(nd_async_device_unregister, dev,
				&nd_async_domain);
		break;
	case ND_SYNC:
570 571 572 573 574 575 576 577 578 579 580 581 582 583
		/*
		 * In the sync case the device is being unregistered due
		 * to a state change of the parent. Claim the kill state
		 * to synchronize against other unregistration requests,
		 * or otherwise let the async path handle it if the
		 * unregistration was already queued.
		 */
		device_lock(dev);
		killed = kill_device(dev);
		device_unlock(dev);

		if (!killed)
			return;

584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607
		nd_synchronize();
		device_unregister(dev);
		break;
	}
}
EXPORT_SYMBOL(nd_device_unregister);

/**
 * __nd_driver_register() - register a region or a namespace driver
 * @nd_drv: driver to register
 * @owner: automatically set by nd_driver_register() macro
 * @mod_name: automatically set by nd_driver_register() macro
 */
int __nd_driver_register(struct nd_device_driver *nd_drv, struct module *owner,
		const char *mod_name)
{
	struct device_driver *drv = &nd_drv->drv;

	if (!nd_drv->type) {
		pr_debug("driver type bitmask not set (%pf)\n",
				__builtin_return_address(0));
		return -EINVAL;
	}

608 609
	if (!nd_drv->probe) {
		pr_debug("%s ->probe() must be specified\n", mod_name);
610 611 612 613 614 615 616 617 618 619 620
		return -EINVAL;
	}

	drv->bus = &nvdimm_bus_type;
	drv->owner = owner;
	drv->mod_name = mod_name;

	return driver_register(drv);
}
EXPORT_SYMBOL(__nd_driver_register);

621 622
int nvdimm_revalidate_disk(struct gendisk *disk)
{
D
Dan Williams 已提交
623
	struct device *dev = disk_to_dev(disk)->parent;
624
	struct nd_region *nd_region = to_nd_region(dev->parent);
625
	int disk_ro = get_disk_ro(disk);
626

627 628 629 630 631
	/*
	 * Upgrade to read-only if the region is read-only preserve as
	 * read-only if the disk is already read-only.
	 */
	if (disk_ro || nd_region->ro == disk_ro)
632 633
		return 0;

634 635 636
	dev_info(dev, "%s read-only, marking %s read-only\n",
			dev_name(&nd_region->dev), disk->disk_name);
	set_disk_ro(disk, 1);
637 638 639 640 641 642

	return 0;

}
EXPORT_SYMBOL(nvdimm_revalidate_disk);

643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663
static ssize_t modalias_show(struct device *dev, struct device_attribute *attr,
		char *buf)
{
	return sprintf(buf, ND_DEVICE_MODALIAS_FMT "\n",
			to_nd_device_type(dev));
}
static DEVICE_ATTR_RO(modalias);

static ssize_t devtype_show(struct device *dev, struct device_attribute *attr,
		char *buf)
{
	return sprintf(buf, "%s\n", dev->type->name);
}
static DEVICE_ATTR_RO(devtype);

static struct attribute *nd_device_attributes[] = {
	&dev_attr_modalias.attr,
	&dev_attr_devtype.attr,
	NULL,
};

664
/*
665 666 667 668
 * nd_device_attribute_group - generic attributes for all devices on an nd bus
 */
struct attribute_group nd_device_attribute_group = {
	.attrs = nd_device_attributes,
669
};
670
EXPORT_SYMBOL_GPL(nd_device_attribute_group);
671

672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692
static ssize_t numa_node_show(struct device *dev,
		struct device_attribute *attr, char *buf)
{
	return sprintf(buf, "%d\n", dev_to_node(dev));
}
static DEVICE_ATTR_RO(numa_node);

static struct attribute *nd_numa_attributes[] = {
	&dev_attr_numa_node.attr,
	NULL,
};

static umode_t nd_numa_attr_visible(struct kobject *kobj, struct attribute *a,
		int n)
{
	if (!IS_ENABLED(CONFIG_NUMA))
		return 0;

	return a->mode;
}

693
/*
694 695 696 697 698 699 700 701
 * nd_numa_attribute_group - NUMA attributes for all devices on an nd bus
 */
struct attribute_group nd_numa_attribute_group = {
	.attrs = nd_numa_attributes,
	.is_visible = nd_numa_attr_visible,
};
EXPORT_SYMBOL_GPL(nd_numa_attribute_group);

702 703 704 705 706 707 708 709
int nvdimm_bus_create_ndctl(struct nvdimm_bus *nvdimm_bus)
{
	dev_t devt = MKDEV(nvdimm_bus_major, nvdimm_bus->id);
	struct device *dev;

	dev = device_create(nd_class, &nvdimm_bus->dev, devt, nvdimm_bus,
			"ndctl%d", nvdimm_bus->id);

D
Dan Williams 已提交
710
	if (IS_ERR(dev))
711 712
		dev_dbg(&nvdimm_bus->dev, "failed to register ndctl%d: %ld\n",
				nvdimm_bus->id, PTR_ERR(dev));
D
Dan Williams 已提交
713
	return PTR_ERR_OR_ZERO(dev);
714 715 716 717 718 719 720
}

void nvdimm_bus_destroy_ndctl(struct nvdimm_bus *nvdimm_bus)
{
	device_destroy(nd_class, MKDEV(nvdimm_bus_major, nvdimm_bus->id));
}

721 722 723 724
static const struct nd_cmd_desc __nd_cmd_dimm_descs[] = {
	[ND_CMD_IMPLEMENTED] = { },
	[ND_CMD_SMART] = {
		.out_num = 2,
725
		.out_sizes = { 4, 128, },
726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756
	},
	[ND_CMD_SMART_THRESHOLD] = {
		.out_num = 2,
		.out_sizes = { 4, 8, },
	},
	[ND_CMD_DIMM_FLAGS] = {
		.out_num = 2,
		.out_sizes = { 4, 4 },
	},
	[ND_CMD_GET_CONFIG_SIZE] = {
		.out_num = 3,
		.out_sizes = { 4, 4, 4, },
	},
	[ND_CMD_GET_CONFIG_DATA] = {
		.in_num = 2,
		.in_sizes = { 4, 4, },
		.out_num = 2,
		.out_sizes = { 4, UINT_MAX, },
	},
	[ND_CMD_SET_CONFIG_DATA] = {
		.in_num = 3,
		.in_sizes = { 4, 4, UINT_MAX, },
		.out_num = 1,
		.out_sizes = { 4, },
	},
	[ND_CMD_VENDOR] = {
		.in_num = 3,
		.in_sizes = { 4, 4, UINT_MAX, },
		.out_num = 3,
		.out_sizes = { 4, 4, UINT_MAX, },
	},
757 758 759 760 761 762
	[ND_CMD_CALL] = {
		.in_num = 2,
		.in_sizes = { sizeof(struct nd_cmd_pkg), UINT_MAX, },
		.out_num = 1,
		.out_sizes = { UINT_MAX, },
	},
763 764 765 766 767 768 769 770 771 772 773 774 775 776 777
};

const struct nd_cmd_desc *nd_cmd_dimm_desc(int cmd)
{
	if (cmd < ARRAY_SIZE(__nd_cmd_dimm_descs))
		return &__nd_cmd_dimm_descs[cmd];
	return NULL;
}
EXPORT_SYMBOL_GPL(nd_cmd_dimm_desc);

static const struct nd_cmd_desc __nd_cmd_bus_descs[] = {
	[ND_CMD_IMPLEMENTED] = { },
	[ND_CMD_ARS_CAP] = {
		.in_num = 2,
		.in_sizes = { 8, 8, },
778 779
		.out_num = 4,
		.out_sizes = { 4, 4, 4, 4, },
780 781
	},
	[ND_CMD_ARS_START] = {
782 783 784 785
		.in_num = 5,
		.in_sizes = { 8, 8, 2, 1, 5, },
		.out_num = 2,
		.out_sizes = { 4, 4, },
786 787
	},
	[ND_CMD_ARS_STATUS] = {
788 789
		.out_num = 3,
		.out_sizes = { 4, 4, UINT_MAX, },
790
	},
791 792 793 794 795 796
	[ND_CMD_CLEAR_ERROR] = {
		.in_num = 2,
		.in_sizes = { 8, 8, },
		.out_num = 3,
		.out_sizes = { 4, 4, 8, },
	},
797 798 799 800 801 802
	[ND_CMD_CALL] = {
		.in_num = 2,
		.in_sizes = { sizeof(struct nd_cmd_pkg), UINT_MAX, },
		.out_num = 1,
		.out_sizes = { UINT_MAX, },
	},
803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829
};

const struct nd_cmd_desc *nd_cmd_bus_desc(int cmd)
{
	if (cmd < ARRAY_SIZE(__nd_cmd_bus_descs))
		return &__nd_cmd_bus_descs[cmd];
	return NULL;
}
EXPORT_SYMBOL_GPL(nd_cmd_bus_desc);

u32 nd_cmd_in_size(struct nvdimm *nvdimm, int cmd,
		const struct nd_cmd_desc *desc, int idx, void *buf)
{
	if (idx >= desc->in_num)
		return UINT_MAX;

	if (desc->in_sizes[idx] < UINT_MAX)
		return desc->in_sizes[idx];

	if (nvdimm && cmd == ND_CMD_SET_CONFIG_DATA && idx == 2) {
		struct nd_cmd_set_config_hdr *hdr = buf;

		return hdr->in_length;
	} else if (nvdimm && cmd == ND_CMD_VENDOR && idx == 2) {
		struct nd_cmd_vendor_hdr *hdr = buf;

		return hdr->in_length;
830 831 832 833
	} else if (cmd == ND_CMD_CALL) {
		struct nd_cmd_pkg *pkg = buf;

		return pkg->nd_size_in;
834 835 836 837 838 839 840 841
	}

	return UINT_MAX;
}
EXPORT_SYMBOL_GPL(nd_cmd_in_size);

u32 nd_cmd_out_size(struct nvdimm *nvdimm, int cmd,
		const struct nd_cmd_desc *desc, int idx, const u32 *in_field,
842
		const u32 *out_field, unsigned long remainder)
843 844 845 846 847 848 849 850 851 852 853
{
	if (idx >= desc->out_num)
		return UINT_MAX;

	if (desc->out_sizes[idx] < UINT_MAX)
		return desc->out_sizes[idx];

	if (nvdimm && cmd == ND_CMD_GET_CONFIG_DATA && idx == 1)
		return in_field[1];
	else if (nvdimm && cmd == ND_CMD_VENDOR && idx == 2)
		return out_field[1];
854 855 856 857 858 859 860 861 862 863 864 865 866 867
	else if (!nvdimm && cmd == ND_CMD_ARS_STATUS && idx == 2) {
		/*
		 * Per table 9-276 ARS Data in ACPI 6.1, out_field[1] is
		 * "Size of Output Buffer in bytes, including this
		 * field."
		 */
		if (out_field[1] < 4)
			return 0;
		/*
		 * ACPI 6.1 is ambiguous if 'status' is included in the
		 * output size. If we encounter an output size that
		 * overshoots the remainder by 4 bytes, assume it was
		 * including 'status'.
		 */
868
		if (out_field[1] - 4 == remainder)
869
			return remainder;
870
		return out_field[1] - 8;
871
	} else if (cmd == ND_CMD_CALL) {
872 873 874 875 876
		struct nd_cmd_pkg *pkg = (struct nd_cmd_pkg *) in_field;

		return pkg->nd_size_out;
	}

877 878 879 880 881

	return UINT_MAX;
}
EXPORT_SYMBOL_GPL(nd_cmd_out_size);

882
void wait_nvdimm_bus_probe_idle(struct device *dev)
883
{
884 885
	struct nvdimm_bus *nvdimm_bus = walk_to_nvdimm_bus(dev);

886 887 888
	do {
		if (nvdimm_bus->probe_active == 0)
			break;
889 890
		nvdimm_bus_unlock(dev);
		device_unlock(dev);
891
		wait_event(nvdimm_bus->wait,
892
				nvdimm_bus->probe_active == 0);
893 894
		device_lock(dev);
		nvdimm_bus_lock(dev);
895 896 897
	} while (true);
}

898
static int nd_pmem_forget_poison_check(struct device *dev, void *data)
899
{
900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932
	struct nd_cmd_clear_error *clear_err =
		(struct nd_cmd_clear_error *)data;
	struct nd_btt *nd_btt = is_nd_btt(dev) ? to_nd_btt(dev) : NULL;
	struct nd_pfn *nd_pfn = is_nd_pfn(dev) ? to_nd_pfn(dev) : NULL;
	struct nd_dax *nd_dax = is_nd_dax(dev) ? to_nd_dax(dev) : NULL;
	struct nd_namespace_common *ndns = NULL;
	struct nd_namespace_io *nsio;
	resource_size_t offset = 0, end_trunc = 0, start, end, pstart, pend;

	if (nd_dax || !dev->driver)
		return 0;

	start = clear_err->address;
	end = clear_err->address + clear_err->cleared - 1;

	if (nd_btt || nd_pfn || nd_dax) {
		if (nd_btt)
			ndns = nd_btt->ndns;
		else if (nd_pfn)
			ndns = nd_pfn->ndns;
		else if (nd_dax)
			ndns = nd_dax->nd_pfn.ndns;

		if (!ndns)
			return 0;
	} else
		ndns = to_ndns(dev);

	nsio = to_nd_namespace_io(&ndns->dev);
	pstart = nsio->res.start + offset;
	pend = nsio->res.end - end_trunc;

	if ((pstart >= start) && (pend <= end))
933
		return -EBUSY;
934

935
	return 0;
936 937 938 939 940 941

}

static int nd_ns_forget_poison_check(struct device *dev, void *data)
{
	return device_for_each_child(dev, data, nd_pmem_forget_poison_check);
942 943
}

944
/* set_config requires an idle interleave set */
945
static int nd_cmd_clear_to_send(struct nvdimm_bus *nvdimm_bus,
946
		struct nvdimm *nvdimm, unsigned int cmd, void *data)
947
{
948 949 950 951
	struct nvdimm_bus_descriptor *nd_desc = nvdimm_bus->nd_desc;

	/* ask the bus provider if it would like to block this request */
	if (nd_desc->clear_to_send) {
952
		int rc = nd_desc->clear_to_send(nd_desc, nvdimm, cmd, data);
953 954 955 956

		if (rc)
			return rc;
	}
957

958 959
	/* require clear error to go through the pmem driver */
	if (!nvdimm && cmd == ND_CMD_CLEAR_ERROR)
960 961
		return device_for_each_child(&nvdimm_bus->dev, data,
				nd_ns_forget_poison_check);
962

963 964 965
	if (!nvdimm || cmd != ND_CMD_SET_CONFIG_DATA)
		return 0;

966
	/* prevent label manipulation while the kernel owns label updates */
967
	wait_nvdimm_bus_probe_idle(&nvdimm_bus->dev);
968 969 970 971 972
	if (atomic_read(&nvdimm->busy))
		return -EBUSY;
	return 0;
}

973 974 975 976 977 978 979
static int __nd_ioctl(struct nvdimm_bus *nvdimm_bus, struct nvdimm *nvdimm,
		int read_only, unsigned int ioctl_cmd, unsigned long arg)
{
	struct nvdimm_bus_descriptor *nd_desc = nvdimm_bus->nd_desc;
	const struct nd_cmd_desc *desc = NULL;
	unsigned int cmd = _IOC_NR(ioctl_cmd);
	struct device *dev = &nvdimm_bus->dev;
980
	void __user *p = (void __user *) arg;
981
	char *out_env = NULL, *in_env = NULL;
982
	const char *cmd_name, *dimm_name;
983 984
	u32 in_len = 0, out_len = 0;
	unsigned int func = cmd;
985
	unsigned long cmd_mask;
986
	struct nd_cmd_pkg pkg;
987
	int rc, i, cmd_rc;
988
	void *buf = NULL;
989
	u64 buf_len = 0;
990 991 992 993

	if (nvdimm) {
		desc = nd_cmd_dimm_desc(cmd);
		cmd_name = nvdimm_cmd_name(cmd);
994
		cmd_mask = nvdimm->cmd_mask;
995 996 997 998
		dimm_name = dev_name(&nvdimm->dev);
	} else {
		desc = nd_cmd_bus_desc(cmd);
		cmd_name = nvdimm_bus_cmd_name(cmd);
999
		cmd_mask = nd_desc->cmd_mask;
1000 1001 1002
		dimm_name = "bus";
	}

1003 1004 1005 1006 1007
	if (cmd == ND_CMD_CALL) {
		if (copy_from_user(&pkg, p, sizeof(pkg)))
			return -EFAULT;
	}

1008 1009 1010 1011
	if (!desc ||
	    (desc->out_num + desc->in_num == 0) ||
	    cmd > ND_CMD_CALL ||
	    !test_bit(cmd, &cmd_mask))
1012 1013 1014 1015
		return -ENOTTY;

	/* fail write commands (when read-only) */
	if (read_only)
1016 1017 1018 1019
		switch (cmd) {
		case ND_CMD_VENDOR:
		case ND_CMD_SET_CONFIG_DATA:
		case ND_CMD_ARS_START:
1020
		case ND_CMD_CLEAR_ERROR:
1021
		case ND_CMD_CALL:
1022
			dev_dbg(dev, "'%s' command while read-only.\n",
1023 1024 1025 1026 1027 1028 1029 1030
					nvdimm ? nvdimm_cmd_name(cmd)
					: nvdimm_bus_cmd_name(cmd));
			return -EPERM;
		default:
			break;
		}

	/* process an input envelope */
1031 1032 1033
	in_env = kzalloc(ND_CMD_MAX_ENVELOPE, GFP_KERNEL);
	if (!in_env)
		return -ENOMEM;
1034 1035 1036 1037 1038 1039 1040
	for (i = 0; i < desc->in_num; i++) {
		u32 in_size, copy;

		in_size = nd_cmd_in_size(nvdimm, cmd, desc, i, in_env);
		if (in_size == UINT_MAX) {
			dev_err(dev, "%s:%s unknown input size cmd: %s field: %d\n",
					__func__, dimm_name, cmd_name, i);
1041 1042
			rc = -ENXIO;
			goto out;
1043
		}
1044 1045
		if (in_len < ND_CMD_MAX_ENVELOPE)
			copy = min_t(u32, ND_CMD_MAX_ENVELOPE - in_len, in_size);
1046 1047
		else
			copy = 0;
1048 1049 1050 1051
		if (copy && copy_from_user(&in_env[in_len], p + in_len, copy)) {
			rc = -EFAULT;
			goto out;
		}
1052 1053 1054
		in_len += in_size;
	}

1055
	if (cmd == ND_CMD_CALL) {
1056
		func = pkg.nd_command;
1057 1058
		dev_dbg(dev, "%s, idx: %llu, in: %u, out: %u, len %llu\n",
				dimm_name, pkg.nd_command,
1059 1060 1061
				in_len, out_len, buf_len);
	}

1062
	/* process an output envelope */
1063 1064 1065 1066 1067 1068
	out_env = kzalloc(ND_CMD_MAX_ENVELOPE, GFP_KERNEL);
	if (!out_env) {
		rc = -ENOMEM;
		goto out;
	}

1069 1070
	for (i = 0; i < desc->out_num; i++) {
		u32 out_size = nd_cmd_out_size(nvdimm, cmd, desc, i,
1071
				(u32 *) in_env, (u32 *) out_env, 0);
1072 1073 1074
		u32 copy;

		if (out_size == UINT_MAX) {
1075 1076
			dev_dbg(dev, "%s unknown output size cmd: %s field: %d\n",
					dimm_name, cmd_name, i);
1077 1078
			rc = -EFAULT;
			goto out;
1079
		}
1080 1081
		if (out_len < ND_CMD_MAX_ENVELOPE)
			copy = min_t(u32, ND_CMD_MAX_ENVELOPE - out_len, out_size);
1082 1083 1084
		else
			copy = 0;
		if (copy && copy_from_user(&out_env[out_len],
1085 1086 1087 1088
					p + in_len + out_len, copy)) {
			rc = -EFAULT;
			goto out;
		}
1089 1090 1091
		out_len += out_size;
	}

1092
	buf_len = (u64) out_len + (u64) in_len;
1093
	if (buf_len > ND_IOCTL_MAX_BUFLEN) {
1094 1095
		dev_dbg(dev, "%s cmd: %s buf_len: %llu > %d\n", dimm_name,
				cmd_name, buf_len, ND_IOCTL_MAX_BUFLEN);
1096 1097
		rc = -EINVAL;
		goto out;
1098 1099 1100
	}

	buf = vmalloc(buf_len);
1101 1102 1103 1104
	if (!buf) {
		rc = -ENOMEM;
		goto out;
	}
1105 1106 1107 1108 1109 1110

	if (copy_from_user(buf, p, buf_len)) {
		rc = -EFAULT;
		goto out;
	}

1111 1112
	device_lock(dev);
	nvdimm_bus_lock(dev);
1113
	rc = nd_cmd_clear_to_send(nvdimm_bus, nvdimm, func, buf);
1114 1115 1116
	if (rc)
		goto out_unlock;

1117
	rc = nd_desc->ndctl(nd_desc, nvdimm, cmd, buf, buf_len, &cmd_rc);
1118
	if (rc < 0)
1119
		goto out_unlock;
1120 1121 1122 1123

	if (!nvdimm && cmd == ND_CMD_CLEAR_ERROR && cmd_rc >= 0) {
		struct nd_cmd_clear_error *clear_err = buf;

1124 1125
		nvdimm_account_cleared_poison(nvdimm_bus, clear_err->address,
				clear_err->cleared);
1126
	}
1127

1128 1129
	if (copy_to_user(p, buf, buf_len))
		rc = -EFAULT;
1130

1131
out_unlock:
1132 1133
	nvdimm_bus_unlock(dev);
	device_unlock(dev);
1134 1135 1136
out:
	kfree(in_env);
	kfree(out_env);
1137 1138 1139 1140
	vfree(buf);
	return rc;
}

1141 1142 1143 1144
enum nd_ioctl_mode {
	BUS_IOCTL,
	DIMM_IOCTL,
};
1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158

static int match_dimm(struct device *dev, void *data)
{
	long id = (long) data;

	if (is_nvdimm(dev)) {
		struct nvdimm *nvdimm = to_nvdimm(dev);

		return nvdimm->id == id;
	}

	return 0;
}

1159 1160 1161
static long nd_ioctl(struct file *file, unsigned int cmd, unsigned long arg,
		enum nd_ioctl_mode mode)

1162
{
1163 1164 1165 1166
	struct nvdimm_bus *nvdimm_bus, *found = NULL;
	long id = (long) file->private_data;
	struct nvdimm *nvdimm = NULL;
	int rc, ro;
1167

1168
	ro = ((file->f_flags & O_ACCMODE) == O_RDONLY);
1169 1170
	mutex_lock(&nvdimm_bus_list_mutex);
	list_for_each_entry(nvdimm_bus, &nvdimm_bus_list, list) {
1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182
		if (mode == DIMM_IOCTL) {
			struct device *dev;

			dev = device_find_child(&nvdimm_bus->dev,
					file->private_data, match_dimm);
			if (!dev)
				continue;
			nvdimm = to_nvdimm(dev);
			found = nvdimm_bus;
		} else if (nvdimm_bus->id == id) {
			found = nvdimm_bus;
		}
1183

1184 1185 1186 1187
		if (found) {
			atomic_inc(&nvdimm_bus->ioctl_active);
			break;
		}
1188 1189 1190
	}
	mutex_unlock(&nvdimm_bus_list_mutex);

1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201
	if (!found)
		return -ENXIO;

	nvdimm_bus = found;
	rc = __nd_ioctl(nvdimm_bus, nvdimm, ro, cmd, arg);

	if (nvdimm)
		put_device(&nvdimm->dev);
	if (atomic_dec_and_test(&nvdimm_bus->ioctl_active))
		wake_up(&nvdimm_bus->wait);

1202 1203 1204
	return rc;
}

1205 1206 1207 1208 1209 1210 1211 1212 1213 1214
static long bus_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
{
	return nd_ioctl(file, cmd, arg, BUS_IOCTL);
}

static long dimm_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
{
	return nd_ioctl(file, cmd, arg, DIMM_IOCTL);
}

1215 1216 1217 1218 1219 1220
static int nd_open(struct inode *inode, struct file *file)
{
	long minor = iminor(inode);

	file->private_data = (void *) minor;
	return 0;
1221 1222 1223 1224
}

static const struct file_operations nvdimm_bus_fops = {
	.owner = THIS_MODULE,
1225
	.open = nd_open,
1226 1227
	.unlocked_ioctl = bus_ioctl,
	.compat_ioctl = bus_ioctl,
1228 1229 1230
	.llseek = noop_llseek,
};

1231 1232 1233
static const struct file_operations nvdimm_fops = {
	.owner = THIS_MODULE,
	.open = nd_open,
1234 1235
	.unlocked_ioctl = dimm_ioctl,
	.compat_ioctl = dimm_ioctl,
1236 1237 1238
	.llseek = noop_llseek,
};

1239 1240 1241 1242
int __init nvdimm_bus_init(void)
{
	int rc;

1243 1244 1245 1246
	rc = bus_register(&nvdimm_bus_type);
	if (rc)
		return rc;

1247 1248
	rc = register_chrdev(0, "ndctl", &nvdimm_bus_fops);
	if (rc < 0)
1249
		goto err_bus_chrdev;
1250 1251
	nvdimm_bus_major = rc;

1252 1253 1254 1255 1256
	rc = register_chrdev(0, "dimmctl", &nvdimm_fops);
	if (rc < 0)
		goto err_dimm_chrdev;
	nvdimm_major = rc;

1257
	nd_class = class_create(THIS_MODULE, "nd");
1258 1259
	if (IS_ERR(nd_class)) {
		rc = PTR_ERR(nd_class);
1260
		goto err_class;
1261
	}
1262

1263 1264 1265 1266
	rc = driver_register(&nd_bus_driver.drv);
	if (rc)
		goto err_nd_bus;

1267 1268
	return 0;

1269 1270
 err_nd_bus:
	class_destroy(nd_class);
1271
 err_class:
1272 1273
	unregister_chrdev(nvdimm_major, "dimmctl");
 err_dimm_chrdev:
1274
	unregister_chrdev(nvdimm_bus_major, "ndctl");
1275
 err_bus_chrdev:
1276
	bus_unregister(&nvdimm_bus_type);
1277 1278 1279 1280

	return rc;
}

1281
void nvdimm_bus_exit(void)
1282
{
1283
	driver_unregister(&nd_bus_driver.drv);
1284 1285
	class_destroy(nd_class);
	unregister_chrdev(nvdimm_bus_major, "ndctl");
1286
	unregister_chrdev(nvdimm_major, "dimmctl");
1287
	bus_unregister(&nvdimm_bus_type);
1288
	ida_destroy(&nd_ida);
1289
}