vfio_iommu_spapr_tce.c 32.8 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
/*
 * VFIO: IOMMU DMA mapping support for TCE on POWER
 *
 * Copyright (C) 2013 IBM Corp.  All rights reserved.
 *     Author: Alexey Kardashevskiy <aik@ozlabs.ru>
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License version 2 as
 * published by the Free Software Foundation.
 *
 * Derived from original vfio_iommu_type1.c:
 * Copyright (C) 2012 Red Hat, Inc.  All rights reserved.
 *     Author: Alex Williamson <alex.williamson@redhat.com>
 */

#include <linux/module.h>
#include <linux/pci.h>
#include <linux/slab.h>
#include <linux/uaccess.h>
#include <linux/err.h>
#include <linux/vfio.h>
22
#include <linux/vmalloc.h>
23 24
#include <asm/iommu.h>
#include <asm/tce.h>
25
#include <asm/mmu_context.h>
26 27 28 29 30 31 32 33

#define DRIVER_VERSION  "0.1"
#define DRIVER_AUTHOR   "aik@ozlabs.ru"
#define DRIVER_DESC     "VFIO IOMMU SPAPR TCE"

static void tce_iommu_detach_group(void *iommu_data,
		struct iommu_group *iommu_group);

34
static long try_increment_locked_vm(struct mm_struct *mm, long npages)
35 36 37
{
	long ret = 0, locked, lock_limit;

38 39
	if (WARN_ON_ONCE(!mm))
		return -EPERM;
40 41 42 43

	if (!npages)
		return 0;

44 45
	down_write(&mm->mmap_sem);
	locked = mm->locked_vm + npages;
46 47 48 49
	lock_limit = rlimit(RLIMIT_MEMLOCK) >> PAGE_SHIFT;
	if (locked > lock_limit && !capable(CAP_IPC_LOCK))
		ret = -ENOMEM;
	else
50
		mm->locked_vm += npages;
51 52 53

	pr_debug("[%d] RLIMIT_MEMLOCK +%ld %ld/%ld%s\n", current->pid,
			npages << PAGE_SHIFT,
54
			mm->locked_vm << PAGE_SHIFT,
55 56 57
			rlimit(RLIMIT_MEMLOCK),
			ret ? " - exceeded" : "");

58
	up_write(&mm->mmap_sem);
59 60 61 62

	return ret;
}

63
static void decrement_locked_vm(struct mm_struct *mm, long npages)
64
{
65 66
	if (!mm || !npages)
		return;
67

68 69 70 71
	down_write(&mm->mmap_sem);
	if (WARN_ON_ONCE(npages > mm->locked_vm))
		npages = mm->locked_vm;
	mm->locked_vm -= npages;
72 73
	pr_debug("[%d] RLIMIT_MEMLOCK -%ld %ld/%ld\n", current->pid,
			npages << PAGE_SHIFT,
74
			mm->locked_vm << PAGE_SHIFT,
75
			rlimit(RLIMIT_MEMLOCK));
76
	up_write(&mm->mmap_sem);
77 78
}

79 80 81 82 83 84 85
/*
 * VFIO IOMMU fd for SPAPR_TCE IOMMU implementation
 *
 * This code handles mapping and unmapping of user data buffers
 * into DMA'ble space using the IOMMU
 */

86 87 88 89 90
struct tce_iommu_group {
	struct list_head next;
	struct iommu_group *grp;
};

91 92 93 94 95 96 97 98 99
/*
 * A container needs to remember which preregistered region  it has
 * referenced to do proper cleanup at the userspace process exit.
 */
struct tce_iommu_prereg {
	struct list_head next;
	struct mm_iommu_table_group_mem_t *mem;
};

100 101 102 103 104 105 106 107
/*
 * The container descriptor supports only a single group per container.
 * Required by the API as the container is not supplied with the IOMMU group
 * at the moment of initialization.
 */
struct tce_container {
	struct mutex lock;
	bool enabled;
108
	bool v2;
109
	bool def_window_pending;
110
	unsigned long locked_pages;
111
	struct mm_struct *mm;
112 113
	struct iommu_table *tables[IOMMU_TABLE_GROUP_MAX_TABLES];
	struct list_head group_list;
114
	struct list_head prereg_list;
115 116
};

117 118 119 120 121 122 123 124 125 126 127 128 129 130
static long tce_iommu_mm_set(struct tce_container *container)
{
	if (container->mm) {
		if (container->mm == current->mm)
			return 0;
		return -EPERM;
	}
	BUG_ON(!current->mm);
	container->mm = current->mm;
	atomic_inc(&container->mm->mm_count);

	return 0;
}

131 132 133 134 135 136 137 138 139 140 141 142 143 144 145
static long tce_iommu_prereg_free(struct tce_container *container,
		struct tce_iommu_prereg *tcemem)
{
	long ret;

	ret = mm_iommu_put(container->mm, tcemem->mem);
	if (ret)
		return ret;

	list_del(&tcemem->next);
	kfree(tcemem);

	return 0;
}

146 147 148 149
static long tce_iommu_unregister_pages(struct tce_container *container,
		__u64 vaddr, __u64 size)
{
	struct mm_iommu_table_group_mem_t *mem;
150 151
	struct tce_iommu_prereg *tcemem;
	bool found = false;
152 153 154 155

	if ((vaddr & ~PAGE_MASK) || (size & ~PAGE_MASK))
		return -EINVAL;

156
	mem = mm_iommu_find(container->mm, vaddr, size >> PAGE_SHIFT);
157 158 159
	if (!mem)
		return -ENOENT;

160 161 162 163 164 165 166 167 168 169 170
	list_for_each_entry(tcemem, &container->prereg_list, next) {
		if (tcemem->mem == mem) {
			found = true;
			break;
		}
	}

	if (!found)
		return -ENOENT;

	return tce_iommu_prereg_free(container, tcemem);
171 172 173 174 175 176 177
}

static long tce_iommu_register_pages(struct tce_container *container,
		__u64 vaddr, __u64 size)
{
	long ret = 0;
	struct mm_iommu_table_group_mem_t *mem = NULL;
178
	struct tce_iommu_prereg *tcemem;
179 180 181 182 183 184
	unsigned long entries = size >> PAGE_SHIFT;

	if ((vaddr & ~PAGE_MASK) || (size & ~PAGE_MASK) ||
			((vaddr + size) < vaddr))
		return -EINVAL;

185 186 187 188 189 190 191 192
	mem = mm_iommu_find(container->mm, vaddr, entries);
	if (mem) {
		list_for_each_entry(tcemem, &container->prereg_list, next) {
			if (tcemem->mem == mem)
				return -EBUSY;
		}
	}

193
	ret = mm_iommu_get(container->mm, vaddr, entries, &mem);
194 195 196
	if (ret)
		return ret;

197 198 199 200
	tcemem = kzalloc(sizeof(*tcemem), GFP_KERNEL);
	tcemem->mem = mem;
	list_add(&tcemem->next, &container->prereg_list);

201 202 203 204 205
	container->enabled = true;

	return 0;
}

206 207
static long tce_iommu_userspace_view_alloc(struct iommu_table *tbl,
		struct mm_struct *mm)
208 209 210 211 212 213 214 215
{
	unsigned long cb = _ALIGN_UP(sizeof(tbl->it_userspace[0]) *
			tbl->it_size, PAGE_SIZE);
	unsigned long *uas;
	long ret;

	BUG_ON(tbl->it_userspace);

216
	ret = try_increment_locked_vm(mm, cb >> PAGE_SHIFT);
217 218 219 220 221
	if (ret)
		return ret;

	uas = vzalloc(cb);
	if (!uas) {
222
		decrement_locked_vm(mm, cb >> PAGE_SHIFT);
223 224 225 226 227 228 229
		return -ENOMEM;
	}
	tbl->it_userspace = uas;

	return 0;
}

230 231
static void tce_iommu_userspace_view_free(struct iommu_table *tbl,
		struct mm_struct *mm)
232 233 234 235 236 237 238 239 240
{
	unsigned long cb = _ALIGN_UP(sizeof(tbl->it_userspace[0]) *
			tbl->it_size, PAGE_SIZE);

	if (!tbl->it_userspace)
		return;

	vfree(tbl->it_userspace);
	tbl->it_userspace = NULL;
241
	decrement_locked_vm(mm, cb >> PAGE_SHIFT);
242 243
}

244 245 246 247 248 249 250 251 252 253
static bool tce_page_is_contained(struct page *page, unsigned page_shift)
{
	/*
	 * Check that the TCE table granularity is not bigger than the size of
	 * a page we just found. Otherwise the hardware can get access to
	 * a bigger memory chunk that it should.
	 */
	return (PAGE_SHIFT + compound_order(compound_head(page))) >= page_shift;
}

254 255 256 257 258
static inline bool tce_groups_attached(struct tce_container *container)
{
	return !list_empty(&container->group_list);
}

259 260 261 262 263 264
static long tce_iommu_find_table(struct tce_container *container,
		phys_addr_t ioba, struct iommu_table **ptbl)
{
	long i;

	for (i = 0; i < IOMMU_TABLE_GROUP_MAX_TABLES; ++i) {
265
		struct iommu_table *tbl = container->tables[i];
266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281

		if (tbl) {
			unsigned long entry = ioba >> tbl->it_page_shift;
			unsigned long start = tbl->it_offset;
			unsigned long end = start + tbl->it_size;

			if ((start <= entry) && (entry < end)) {
				*ptbl = tbl;
				return i;
			}
		}
	}

	return -1;
}

282 283 284 285 286 287 288 289 290 291 292 293
static int tce_iommu_find_free_table(struct tce_container *container)
{
	int i;

	for (i = 0; i < IOMMU_TABLE_GROUP_MAX_TABLES; ++i) {
		if (!container->tables[i])
			return i;
	}

	return -ENOSPC;
}

294 295 296
static int tce_iommu_enable(struct tce_container *container)
{
	int ret = 0;
297
	unsigned long locked;
298
	struct iommu_table_group *table_group;
299
	struct tce_iommu_group *tcegrp;
300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320

	if (container->enabled)
		return -EBUSY;

	/*
	 * When userspace pages are mapped into the IOMMU, they are effectively
	 * locked memory, so, theoretically, we need to update the accounting
	 * of locked pages on each map and unmap.  For powerpc, the map unmap
	 * paths can be very hot, though, and the accounting would kill
	 * performance, especially since it would be difficult to impossible
	 * to handle the accounting in real mode only.
	 *
	 * To address that, rather than precisely accounting every page, we
	 * instead account for a worst case on locked memory when the iommu is
	 * enabled and disabled.  The worst case upper bound on locked memory
	 * is the size of the whole iommu window, which is usually relatively
	 * small (compared to total memory sizes) on POWER hardware.
	 *
	 * Also we don't have a nice way to fail on H_PUT_TCE due to ulimits,
	 * that would effectively kill the guest at random points, much better
	 * enforcing the limit based on the max that the guest can map.
321 322 323 324 325 326 327
	 *
	 * Unfortunately at the moment it counts whole tables, no matter how
	 * much memory the guest has. I.e. for 4GB guest and 4 IOMMU groups
	 * each with 2GB DMA window, 8GB will be counted here. The reason for
	 * this is that we cannot tell here the amount of RAM used by the guest
	 * as this information is only available from KVM and VFIO is
	 * KVM agnostic.
328 329 330 331
	 *
	 * So we do not allow enabling a container without a group attached
	 * as there is no way to know how much we should increment
	 * the locked_vm counter.
332
	 */
333 334 335 336 337 338
	if (!tce_groups_attached(container))
		return -ENODEV;

	tcegrp = list_first_entry(&container->group_list,
			struct tce_iommu_group, next);
	table_group = iommu_group_get_iommudata(tcegrp->grp);
339 340 341
	if (!table_group)
		return -ENODEV;

342 343 344
	if (!table_group->tce32_size)
		return -EPERM;

345 346 347 348
	ret = tce_iommu_mm_set(container);
	if (ret)
		return ret;

349
	locked = table_group->tce32_size >> PAGE_SHIFT;
350
	ret = try_increment_locked_vm(container->mm, locked);
351 352
	if (ret)
		return ret;
353

354 355 356
	container->locked_pages = locked;

	container->enabled = true;
357 358 359 360 361 362 363 364 365 366 367

	return ret;
}

static void tce_iommu_disable(struct tce_container *container)
{
	if (!container->enabled)
		return;

	container->enabled = false;

368 369
	BUG_ON(!container->mm);
	decrement_locked_vm(container->mm, container->locked_pages);
370 371 372 373 374 375
}

static void *tce_iommu_open(unsigned long arg)
{
	struct tce_container *container;

376
	if ((arg != VFIO_SPAPR_TCE_IOMMU) && (arg != VFIO_SPAPR_TCE_v2_IOMMU)) {
377 378 379 380 381 382 383 384 385
		pr_err("tce_vfio: Wrong IOMMU type\n");
		return ERR_PTR(-EINVAL);
	}

	container = kzalloc(sizeof(*container), GFP_KERNEL);
	if (!container)
		return ERR_PTR(-ENOMEM);

	mutex_init(&container->lock);
386
	INIT_LIST_HEAD_RCU(&container->group_list);
387
	INIT_LIST_HEAD_RCU(&container->prereg_list);
388 389

	container->v2 = arg == VFIO_SPAPR_TCE_v2_IOMMU;
390 391 392 393

	return container;
}

394 395 396
static int tce_iommu_clear(struct tce_container *container,
		struct iommu_table *tbl,
		unsigned long entry, unsigned long pages);
397 398
static void tce_iommu_free_table(struct tce_container *container,
		struct iommu_table *tbl);
399

400 401 402
static void tce_iommu_release(void *iommu_data)
{
	struct tce_container *container = iommu_data;
403 404
	struct tce_iommu_group *tcegrp;
	long i;
405

406 407 408 409 410
	while (tce_groups_attached(container)) {
		tcegrp = list_first_entry(&container->group_list,
				struct tce_iommu_group, next);
		tce_iommu_detach_group(iommu_data, tcegrp->grp);
	}
411

412 413 414 415 416 417 418 419 420 421 422
	/*
	 * If VFIO created a table, it was not disposed
	 * by tce_iommu_detach_group() so do it now.
	 */
	for (i = 0; i < IOMMU_TABLE_GROUP_MAX_TABLES; ++i) {
		struct iommu_table *tbl = container->tables[i];

		if (!tbl)
			continue;

		tce_iommu_clear(container, tbl, tbl->it_offset, tbl->it_size);
423
		tce_iommu_free_table(container, tbl);
424
	}
425

426 427 428 429 430 431 432 433
	while (!list_empty(&container->prereg_list)) {
		struct tce_iommu_prereg *tcemem;

		tcemem = list_first_entry(&container->prereg_list,
				struct tce_iommu_prereg, next);
		WARN_ON_ONCE(tce_iommu_prereg_free(container, tcemem));
	}

434
	tce_iommu_disable(container);
435 436
	if (container->mm)
		mmdrop(container->mm);
437 438 439 440 441
	mutex_destroy(&container->lock);

	kfree(container);
}

442
static void tce_iommu_unuse_page(struct tce_container *container,
443
		unsigned long hpa)
444 445 446
{
	struct page *page;

447
	page = pfn_to_page(hpa >> PAGE_SHIFT);
448 449 450
	put_page(page);
}

451 452
static int tce_iommu_prereg_ua_to_hpa(struct tce_container *container,
		unsigned long tce, unsigned long size,
453 454 455 456 457
		unsigned long *phpa, struct mm_iommu_table_group_mem_t **pmem)
{
	long ret = 0;
	struct mm_iommu_table_group_mem_t *mem;

458
	mem = mm_iommu_lookup(container->mm, tce, size);
459 460 461 462 463 464 465 466 467 468 469 470
	if (!mem)
		return -EINVAL;

	ret = mm_iommu_ua_to_hpa(mem, tce, phpa);
	if (ret)
		return -EINVAL;

	*pmem = mem;

	return 0;
}

471 472
static void tce_iommu_unuse_page_v2(struct tce_container *container,
		struct iommu_table *tbl, unsigned long entry)
473 474 475 476 477 478
{
	struct mm_iommu_table_group_mem_t *mem = NULL;
	int ret;
	unsigned long hpa = 0;
	unsigned long *pua = IOMMU_TABLE_USERSPACE_ENTRY(tbl, entry);

479
	if (!pua)
480 481
		return;

482
	ret = tce_iommu_prereg_ua_to_hpa(container, *pua, IOMMU_PAGE_SIZE(tbl),
483 484 485 486 487 488 489 490 491 492
			&hpa, &mem);
	if (ret)
		pr_debug("%s: tce %lx at #%lx was not cached, ret=%d\n",
				__func__, *pua, entry, ret);
	if (mem)
		mm_iommu_mapped_dec(mem);

	*pua = 0;
}

493 494 495 496
static int tce_iommu_clear(struct tce_container *container,
		struct iommu_table *tbl,
		unsigned long entry, unsigned long pages)
{
497 498 499
	unsigned long oldhpa;
	long ret;
	enum dma_data_direction direction;
500 501

	for ( ; pages; --pages, ++entry) {
502 503 504 505 506 507 508
		direction = DMA_NONE;
		oldhpa = 0;
		ret = iommu_tce_xchg(tbl, entry, &oldhpa, &direction);
		if (ret)
			continue;

		if (direction == DMA_NONE)
509 510
			continue;

511
		if (container->v2) {
512
			tce_iommu_unuse_page_v2(container, tbl, entry);
513 514 515
			continue;
		}

516
		tce_iommu_unuse_page(container, oldhpa);
517 518 519 520 521
	}

	return 0;
}

522 523 524 525 526 527 528 529 530 531 532 533 534 535
static int tce_iommu_use_page(unsigned long tce, unsigned long *hpa)
{
	struct page *page = NULL;
	enum dma_data_direction direction = iommu_tce_direction(tce);

	if (get_user_pages_fast(tce & PAGE_MASK, 1,
			direction != DMA_TO_DEVICE, &page) != 1)
		return -EFAULT;

	*hpa = __pa((unsigned long) page_address(page));

	return 0;
}

536 537
static long tce_iommu_build(struct tce_container *container,
		struct iommu_table *tbl,
538 539
		unsigned long entry, unsigned long tce, unsigned long pages,
		enum dma_data_direction direction)
540 541
{
	long i, ret = 0;
542 543
	struct page *page;
	unsigned long hpa;
544
	enum dma_data_direction dirtmp;
545 546 547 548

	for (i = 0; i < pages; ++i) {
		unsigned long offset = tce & IOMMU_PAGE_MASK(tbl) & ~PAGE_MASK;

549 550
		ret = tce_iommu_use_page(tce, &hpa);
		if (ret)
551
			break;
552

553
		page = pfn_to_page(hpa >> PAGE_SHIFT);
554 555 556 557 558
		if (!tce_page_is_contained(page, tbl->it_page_shift)) {
			ret = -EPERM;
			break;
		}

559
		hpa |= offset;
560 561
		dirtmp = direction;
		ret = iommu_tce_xchg(tbl, entry + i, &hpa, &dirtmp);
562
		if (ret) {
563
			tce_iommu_unuse_page(container, hpa);
564 565 566 567 568
			pr_err("iommu_tce: %s failed ioba=%lx, tce=%lx, ret=%ld\n",
					__func__, entry << tbl->it_page_shift,
					tce, ret);
			break;
		}
569 570 571 572

		if (dirtmp != DMA_NONE)
			tce_iommu_unuse_page(container, hpa);

573
		tce += IOMMU_PAGE_SIZE(tbl);
574 575 576 577 578 579 580 581
	}

	if (ret)
		tce_iommu_clear(container, tbl, entry, i);

	return ret;
}

582 583 584 585 586 587 588 589 590 591
static long tce_iommu_build_v2(struct tce_container *container,
		struct iommu_table *tbl,
		unsigned long entry, unsigned long tce, unsigned long pages,
		enum dma_data_direction direction)
{
	long i, ret = 0;
	struct page *page;
	unsigned long hpa;
	enum dma_data_direction dirtmp;

592
	if (!tbl->it_userspace) {
593
		ret = tce_iommu_userspace_view_alloc(tbl, container->mm);
594 595 596 597
		if (ret)
			return ret;
	}

598 599 600 601 602
	for (i = 0; i < pages; ++i) {
		struct mm_iommu_table_group_mem_t *mem = NULL;
		unsigned long *pua = IOMMU_TABLE_USERSPACE_ENTRY(tbl,
				entry + i);

603 604
		ret = tce_iommu_prereg_ua_to_hpa(container,
				tce, IOMMU_PAGE_SIZE(tbl), &hpa, &mem);
605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624
		if (ret)
			break;

		page = pfn_to_page(hpa >> PAGE_SHIFT);
		if (!tce_page_is_contained(page, tbl->it_page_shift)) {
			ret = -EPERM;
			break;
		}

		/* Preserve offset within IOMMU page */
		hpa |= tce & IOMMU_PAGE_MASK(tbl) & ~PAGE_MASK;
		dirtmp = direction;

		/* The registered region is being unregistered */
		if (mm_iommu_mapped_inc(mem))
			break;

		ret = iommu_tce_xchg(tbl, entry + i, &hpa, &dirtmp);
		if (ret) {
			/* dirtmp cannot be DMA_NONE here */
625
			tce_iommu_unuse_page_v2(container, tbl, entry + i);
626 627 628 629 630 631 632
			pr_err("iommu_tce: %s failed ioba=%lx, tce=%lx, ret=%ld\n",
					__func__, entry << tbl->it_page_shift,
					tce, ret);
			break;
		}

		if (dirtmp != DMA_NONE)
633
			tce_iommu_unuse_page_v2(container, tbl, entry + i);
634 635 636 637 638 639 640 641 642 643 644 645

		*pua = tce;

		tce += IOMMU_PAGE_SIZE(tbl);
	}

	if (ret)
		tce_iommu_clear(container, tbl, entry, i);

	return ret;
}

646 647 648 649 650 651 652 653 654 655 656 657 658 659 660
static long tce_iommu_create_table(struct tce_container *container,
			struct iommu_table_group *table_group,
			int num,
			__u32 page_shift,
			__u64 window_size,
			__u32 levels,
			struct iommu_table **ptbl)
{
	long ret, table_size;

	table_size = table_group->ops->get_table_size(page_shift, window_size,
			levels);
	if (!table_size)
		return -EINVAL;

661
	ret = try_increment_locked_vm(container->mm, table_size >> PAGE_SHIFT);
662 663 664 665 666 667 668 669 670 671 672 673
	if (ret)
		return ret;

	ret = table_group->ops->create_table(table_group, num,
			page_shift, window_size, levels, ptbl);

	WARN_ON(!ret && !(*ptbl)->it_ops->free);
	WARN_ON(!ret && ((*ptbl)->it_allocated_size != table_size));

	return ret;
}

674 675
static void tce_iommu_free_table(struct tce_container *container,
		struct iommu_table *tbl)
676 677 678
{
	unsigned long pages = tbl->it_allocated_size >> PAGE_SHIFT;

679
	tce_iommu_userspace_view_free(tbl, container->mm);
680
	tbl->it_ops->free(tbl);
681
	decrement_locked_vm(container->mm, pages);
682 683
}

684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743
static long tce_iommu_create_window(struct tce_container *container,
		__u32 page_shift, __u64 window_size, __u32 levels,
		__u64 *start_addr)
{
	struct tce_iommu_group *tcegrp;
	struct iommu_table_group *table_group;
	struct iommu_table *tbl = NULL;
	long ret, num;

	num = tce_iommu_find_free_table(container);
	if (num < 0)
		return num;

	/* Get the first group for ops::create_table */
	tcegrp = list_first_entry(&container->group_list,
			struct tce_iommu_group, next);
	table_group = iommu_group_get_iommudata(tcegrp->grp);
	if (!table_group)
		return -EFAULT;

	if (!(table_group->pgsizes & (1ULL << page_shift)))
		return -EINVAL;

	if (!table_group->ops->set_window || !table_group->ops->unset_window ||
			!table_group->ops->get_table_size ||
			!table_group->ops->create_table)
		return -EPERM;

	/* Create TCE table */
	ret = tce_iommu_create_table(container, table_group, num,
			page_shift, window_size, levels, &tbl);
	if (ret)
		return ret;

	BUG_ON(!tbl->it_ops->free);

	/*
	 * Program the table to every group.
	 * Groups have been tested for compatibility at the attach time.
	 */
	list_for_each_entry(tcegrp, &container->group_list, next) {
		table_group = iommu_group_get_iommudata(tcegrp->grp);

		ret = table_group->ops->set_window(table_group, num, tbl);
		if (ret)
			goto unset_exit;
	}

	container->tables[num] = tbl;

	/* Return start address assigned by platform in create_table() */
	*start_addr = tbl->it_offset << tbl->it_page_shift;

	return 0;

unset_exit:
	list_for_each_entry(tcegrp, &container->group_list, next) {
		table_group = iommu_group_get_iommudata(tcegrp->grp);
		table_group->ops->unset_window(table_group, num);
	}
744
	tce_iommu_free_table(container, tbl);
745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781

	return ret;
}

static long tce_iommu_remove_window(struct tce_container *container,
		__u64 start_addr)
{
	struct iommu_table_group *table_group = NULL;
	struct iommu_table *tbl;
	struct tce_iommu_group *tcegrp;
	int num;

	num = tce_iommu_find_table(container, start_addr, &tbl);
	if (num < 0)
		return -EINVAL;

	BUG_ON(!tbl->it_size);

	/* Detach groups from IOMMUs */
	list_for_each_entry(tcegrp, &container->group_list, next) {
		table_group = iommu_group_get_iommudata(tcegrp->grp);

		/*
		 * SPAPR TCE IOMMU exposes the default DMA window to
		 * the guest via dma32_window_start/size of
		 * VFIO_IOMMU_SPAPR_TCE_GET_INFO. Some platforms allow
		 * the userspace to remove this window, some do not so
		 * here we check for the platform capability.
		 */
		if (!table_group->ops || !table_group->ops->unset_window)
			return -EPERM;

		table_group->ops->unset_window(table_group, num);
	}

	/* Free table */
	tce_iommu_clear(container, tbl, tbl->it_offset, tbl->it_size);
782
	tce_iommu_free_table(container, tbl);
783 784 785 786 787
	container->tables[num] = NULL;

	return 0;
}

788 789 790 791 792 793 794
static long tce_iommu_create_default_window(struct tce_container *container)
{
	long ret;
	__u64 start_addr = 0;
	struct tce_iommu_group *tcegrp;
	struct iommu_table_group *table_group;

795 796 797
	if (!container->def_window_pending)
		return 0;

798 799 800 801 802 803 804 805 806 807 808 809 810
	if (!tce_groups_attached(container))
		return -ENODEV;

	tcegrp = list_first_entry(&container->group_list,
			struct tce_iommu_group, next);
	table_group = iommu_group_get_iommudata(tcegrp->grp);
	if (!table_group)
		return -ENODEV;

	ret = tce_iommu_create_window(container, IOMMU_PAGE_SHIFT_4K,
			table_group->tce32_size, 1, &start_addr);
	WARN_ON_ONCE(!ret && start_addr);

811 812 813
	if (!ret)
		container->def_window_pending = false;

814 815 816
	return ret;
}

817 818 819 820
static long tce_iommu_ioctl(void *iommu_data,
				 unsigned int cmd, unsigned long arg)
{
	struct tce_container *container = iommu_data;
821
	unsigned long minsz, ddwsz;
822 823 824 825
	long ret;

	switch (cmd) {
	case VFIO_CHECK_EXTENSION:
826 827
		switch (arg) {
		case VFIO_SPAPR_TCE_IOMMU:
828
		case VFIO_SPAPR_TCE_v2_IOMMU:
829 830 831 832 833 834 835 836
			ret = 1;
			break;
		default:
			ret = vfio_spapr_iommu_eeh_ioctl(NULL, cmd, arg);
			break;
		}

		return (ret < 0) ? 0 : ret;
837 838 839 840 841 842 843 844 845
	}

	/*
	 * Sanity check to prevent one userspace from manipulating
	 * another userspace mm.
	 */
	BUG_ON(!container);
	if (container->mm && container->mm != current->mm)
		return -EPERM;
846

847
	switch (cmd) {
848 849
	case VFIO_IOMMU_SPAPR_TCE_GET_INFO: {
		struct vfio_iommu_spapr_tce_info info;
850
		struct tce_iommu_group *tcegrp;
851 852
		struct iommu_table_group *table_group;

853
		if (!tce_groups_attached(container))
854 855
			return -ENXIO;

856 857 858
		tcegrp = list_first_entry(&container->group_list,
				struct tce_iommu_group, next);
		table_group = iommu_group_get_iommudata(tcegrp->grp);
859

860
		if (!table_group)
861 862 863 864 865 866 867 868 869 870 871
			return -ENXIO;

		minsz = offsetofend(struct vfio_iommu_spapr_tce_info,
				dma32_window_size);

		if (copy_from_user(&info, (void __user *)arg, minsz))
			return -EFAULT;

		if (info.argsz < minsz)
			return -EINVAL;

872 873
		info.dma32_window_start = table_group->tce32_start;
		info.dma32_window_size = table_group->tce32_size;
874
		info.flags = 0;
875 876 877 878 879 880 881 882 883 884 885 886 887 888 889
		memset(&info.ddw, 0, sizeof(info.ddw));

		if (table_group->max_dynamic_windows_supported &&
				container->v2) {
			info.flags |= VFIO_IOMMU_SPAPR_INFO_DDW;
			info.ddw.pgsizes = table_group->pgsizes;
			info.ddw.max_dynamic_windows_supported =
				table_group->max_dynamic_windows_supported;
			info.ddw.levels = table_group->max_levels;
		}

		ddwsz = offsetofend(struct vfio_iommu_spapr_tce_info, ddw);

		if (info.argsz >= ddwsz)
			minsz = ddwsz;
890 891 892 893 894 895 896 897

		if (copy_to_user((void __user *)arg, &info, minsz))
			return -EFAULT;

		return 0;
	}
	case VFIO_IOMMU_MAP_DMA: {
		struct vfio_iommu_type1_dma_map param;
898 899
		struct iommu_table *tbl = NULL;
		long num;
900
		enum dma_data_direction direction;
901

902 903 904
		if (!container->enabled)
			return -EPERM;

905 906 907 908 909 910 911 912 913 914 915 916
		minsz = offsetofend(struct vfio_iommu_type1_dma_map, size);

		if (copy_from_user(&param, (void __user *)arg, minsz))
			return -EFAULT;

		if (param.argsz < minsz)
			return -EINVAL;

		if (param.flags & ~(VFIO_DMA_MAP_FLAG_READ |
				VFIO_DMA_MAP_FLAG_WRITE))
			return -EINVAL;

917 918 919 920
		ret = tce_iommu_create_default_window(container);
		if (ret)
			return ret;

921 922 923 924
		num = tce_iommu_find_table(container, param.iova, &tbl);
		if (num < 0)
			return -ENXIO;

925 926
		if ((param.size & ~IOMMU_PAGE_MASK(tbl)) ||
				(param.vaddr & ~IOMMU_PAGE_MASK(tbl)))
927 928 929
			return -EINVAL;

		/* iova is checked by the IOMMU API */
930 931 932 933 934 935 936 937 938 939 940
		if (param.flags & VFIO_DMA_MAP_FLAG_READ) {
			if (param.flags & VFIO_DMA_MAP_FLAG_WRITE)
				direction = DMA_BIDIRECTIONAL;
			else
				direction = DMA_TO_DEVICE;
		} else {
			if (param.flags & VFIO_DMA_MAP_FLAG_WRITE)
				direction = DMA_FROM_DEVICE;
			else
				return -EINVAL;
		}
941

942
		ret = iommu_tce_put_param_check(tbl, param.iova, param.vaddr);
943 944 945
		if (ret)
			return ret;

946 947 948 949 950 951 952 953 954 955 956 957
		if (container->v2)
			ret = tce_iommu_build_v2(container, tbl,
					param.iova >> tbl->it_page_shift,
					param.vaddr,
					param.size >> tbl->it_page_shift,
					direction);
		else
			ret = tce_iommu_build(container, tbl,
					param.iova >> tbl->it_page_shift,
					param.vaddr,
					param.size >> tbl->it_page_shift,
					direction);
958 959 960 961 962 963 964

		iommu_flush_tce(tbl);

		return ret;
	}
	case VFIO_IOMMU_UNMAP_DMA: {
		struct vfio_iommu_type1_dma_unmap param;
965 966
		struct iommu_table *tbl = NULL;
		long num;
967

968 969 970
		if (!container->enabled)
			return -EPERM;

971 972 973 974 975 976 977 978 979 980 981 982 983
		minsz = offsetofend(struct vfio_iommu_type1_dma_unmap,
				size);

		if (copy_from_user(&param, (void __user *)arg, minsz))
			return -EFAULT;

		if (param.argsz < minsz)
			return -EINVAL;

		/* No flag is supported now */
		if (param.flags)
			return -EINVAL;

984 985 986 987
		ret = tce_iommu_create_default_window(container);
		if (ret)
			return ret;

988 989 990 991
		num = tce_iommu_find_table(container, param.iova, &tbl);
		if (num < 0)
			return -ENXIO;

992
		if (param.size & ~IOMMU_PAGE_MASK(tbl))
993 994 995
			return -EINVAL;

		ret = iommu_tce_clear_param_check(tbl, param.iova, 0,
996
				param.size >> tbl->it_page_shift);
997 998 999
		if (ret)
			return ret;

1000
		ret = tce_iommu_clear(container, tbl,
1001 1002
				param.iova >> tbl->it_page_shift,
				param.size >> tbl->it_page_shift);
1003 1004 1005 1006
		iommu_flush_tce(tbl);

		return ret;
	}
1007 1008 1009 1010 1011 1012 1013 1014 1015
	case VFIO_IOMMU_SPAPR_REGISTER_MEMORY: {
		struct vfio_iommu_spapr_register_memory param;

		if (!container->v2)
			break;

		minsz = offsetofend(struct vfio_iommu_spapr_register_memory,
				size);

1016 1017 1018 1019
		ret = tce_iommu_mm_set(container);
		if (ret)
			return ret;

1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042
		if (copy_from_user(&param, (void __user *)arg, minsz))
			return -EFAULT;

		if (param.argsz < minsz)
			return -EINVAL;

		/* No flag is supported now */
		if (param.flags)
			return -EINVAL;

		mutex_lock(&container->lock);
		ret = tce_iommu_register_pages(container, param.vaddr,
				param.size);
		mutex_unlock(&container->lock);

		return ret;
	}
	case VFIO_IOMMU_SPAPR_UNREGISTER_MEMORY: {
		struct vfio_iommu_spapr_register_memory param;

		if (!container->v2)
			break;

1043 1044 1045
		if (!container->mm)
			return -EPERM;

1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065
		minsz = offsetofend(struct vfio_iommu_spapr_register_memory,
				size);

		if (copy_from_user(&param, (void __user *)arg, minsz))
			return -EFAULT;

		if (param.argsz < minsz)
			return -EINVAL;

		/* No flag is supported now */
		if (param.flags)
			return -EINVAL;

		mutex_lock(&container->lock);
		ret = tce_iommu_unregister_pages(container, param.vaddr,
				param.size);
		mutex_unlock(&container->lock);

		return ret;
	}
1066
	case VFIO_IOMMU_ENABLE:
1067 1068 1069
		if (container->v2)
			break;

1070 1071 1072 1073 1074 1075 1076
		mutex_lock(&container->lock);
		ret = tce_iommu_enable(container);
		mutex_unlock(&container->lock);
		return ret;


	case VFIO_IOMMU_DISABLE:
1077 1078 1079
		if (container->v2)
			break;

1080 1081 1082 1083
		mutex_lock(&container->lock);
		tce_iommu_disable(container);
		mutex_unlock(&container->lock);
		return 0;
1084

1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097
	case VFIO_EEH_PE_OP: {
		struct tce_iommu_group *tcegrp;

		ret = 0;
		list_for_each_entry(tcegrp, &container->group_list, next) {
			ret = vfio_spapr_iommu_eeh_ioctl(tcegrp->grp,
					cmd, arg);
			if (ret)
				return ret;
		}
		return ret;
	}

1098 1099 1100 1101 1102 1103
	case VFIO_IOMMU_SPAPR_TCE_CREATE: {
		struct vfio_iommu_spapr_tce_create create;

		if (!container->v2)
			break;

1104 1105 1106 1107
		ret = tce_iommu_mm_set(container);
		if (ret)
			return ret;

1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124
		if (!tce_groups_attached(container))
			return -ENXIO;

		minsz = offsetofend(struct vfio_iommu_spapr_tce_create,
				start_addr);

		if (copy_from_user(&create, (void __user *)arg, minsz))
			return -EFAULT;

		if (create.argsz < minsz)
			return -EINVAL;

		if (create.flags)
			return -EINVAL;

		mutex_lock(&container->lock);

1125 1126 1127 1128
		ret = tce_iommu_create_default_window(container);
		if (ret)
			return ret;

1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145
		ret = tce_iommu_create_window(container, create.page_shift,
				create.window_size, create.levels,
				&create.start_addr);

		mutex_unlock(&container->lock);

		if (!ret && copy_to_user((void __user *)arg, &create, minsz))
			ret = -EFAULT;

		return ret;
	}
	case VFIO_IOMMU_SPAPR_TCE_REMOVE: {
		struct vfio_iommu_spapr_tce_remove remove;

		if (!container->v2)
			break;

1146 1147 1148 1149
		ret = tce_iommu_mm_set(container);
		if (ret)
			return ret;

1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164
		if (!tce_groups_attached(container))
			return -ENXIO;

		minsz = offsetofend(struct vfio_iommu_spapr_tce_remove,
				start_addr);

		if (copy_from_user(&remove, (void __user *)arg, minsz))
			return -EFAULT;

		if (remove.argsz < minsz)
			return -EINVAL;

		if (remove.flags)
			return -EINVAL;

1165 1166 1167 1168 1169
		if (container->def_window_pending && !remove.start_addr) {
			container->def_window_pending = false;
			return 0;
		}

1170 1171 1172 1173 1174 1175 1176 1177
		mutex_lock(&container->lock);

		ret = tce_iommu_remove_window(container, remove.start_addr);

		mutex_unlock(&container->lock);

		return ret;
	}
1178 1179 1180 1181 1182
	}

	return -ENOTTY;
}

1183 1184 1185 1186 1187 1188
static void tce_iommu_release_ownership(struct tce_container *container,
		struct iommu_table_group *table_group)
{
	int i;

	for (i = 0; i < IOMMU_TABLE_GROUP_MAX_TABLES; ++i) {
1189
		struct iommu_table *tbl = container->tables[i];
1190 1191 1192 1193 1194

		if (!tbl)
			continue;

		tce_iommu_clear(container, tbl, tbl->it_offset, tbl->it_size);
1195
		tce_iommu_userspace_view_free(tbl, container->mm);
1196 1197
		if (tbl->it_map)
			iommu_release_ownership(tbl);
1198 1199

		container->tables[i] = NULL;
1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213
	}
}

static int tce_iommu_take_ownership(struct tce_container *container,
		struct iommu_table_group *table_group)
{
	int i, j, rc = 0;

	for (i = 0; i < IOMMU_TABLE_GROUP_MAX_TABLES; ++i) {
		struct iommu_table *tbl = table_group->tables[i];

		if (!tbl || !tbl->it_map)
			continue;

1214
		rc = iommu_take_ownership(tbl);
1215 1216 1217 1218 1219 1220 1221 1222 1223
		if (rc) {
			for (j = 0; j < i; ++j)
				iommu_release_ownership(
						table_group->tables[j]);

			return rc;
		}
	}

1224 1225 1226
	for (i = 0; i < IOMMU_TABLE_GROUP_MAX_TABLES; ++i)
		container->tables[i] = table_group->tables[i];

1227 1228 1229 1230 1231 1232
	return 0;
}

static void tce_iommu_release_ownership_ddw(struct tce_container *container,
		struct iommu_table_group *table_group)
{
1233 1234 1235 1236 1237 1238 1239
	long i;

	if (!table_group->ops->unset_window) {
		WARN_ON_ONCE(1);
		return;
	}

1240
	for (i = 0; i < IOMMU_TABLE_GROUP_MAX_TABLES; ++i)
1241 1242
		table_group->ops->unset_window(table_group, i);

1243 1244 1245 1246 1247 1248
	table_group->ops->release_ownership(table_group);
}

static long tce_iommu_take_ownership_ddw(struct tce_container *container,
		struct iommu_table_group *table_group)
{
1249 1250
	long i, ret = 0;

1251 1252 1253 1254 1255 1256
	if (!table_group->ops->create_table || !table_group->ops->set_window ||
			!table_group->ops->release_ownership) {
		WARN_ON_ONCE(1);
		return -EFAULT;
	}

1257 1258
	table_group->ops->take_ownership(table_group);

1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270
	/* Set all windows to the new group */
	for (i = 0; i < IOMMU_TABLE_GROUP_MAX_TABLES; ++i) {
		struct iommu_table *tbl = container->tables[i];

		if (!tbl)
			continue;

		ret = table_group->ops->set_window(table_group, i, tbl);
		if (ret)
			goto release_exit;
	}

1271
	return 0;
1272 1273 1274 1275 1276 1277 1278 1279

release_exit:
	for (i = 0; i < IOMMU_TABLE_GROUP_MAX_TABLES; ++i)
		table_group->ops->unset_window(table_group, i);

	table_group->ops->release_ownership(table_group);

	return ret;
1280 1281
}

1282 1283 1284 1285 1286
static int tce_iommu_attach_group(void *iommu_data,
		struct iommu_group *iommu_group)
{
	int ret;
	struct tce_container *container = iommu_data;
1287
	struct iommu_table_group *table_group;
1288
	struct tce_iommu_group *tcegrp = NULL;
1289 1290 1291 1292 1293

	mutex_lock(&container->lock);

	/* pr_debug("tce_vfio: Attaching group #%u to iommu %p\n",
			iommu_group_id(iommu_group), iommu_group); */
1294
	table_group = iommu_group_get_iommudata(iommu_group);
1295 1296 1297 1298
	if (!table_group) {
		ret = -ENODEV;
		goto unlock_exit;
	}
1299 1300 1301 1302

	if (tce_groups_attached(container) && (!table_group->ops ||
			!table_group->ops->take_ownership ||
			!table_group->ops->release_ownership)) {
1303
		ret = -EBUSY;
1304 1305 1306
		goto unlock_exit;
	}

1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317
	/* Check if new group has the same iommu_ops (i.e. compatible) */
	list_for_each_entry(tcegrp, &container->group_list, next) {
		struct iommu_table_group *table_group_tmp;

		if (tcegrp->grp == iommu_group) {
			pr_warn("tce_vfio: Group %d is already attached\n",
					iommu_group_id(iommu_group));
			ret = -EBUSY;
			goto unlock_exit;
		}
		table_group_tmp = iommu_group_get_iommudata(tcegrp->grp);
1318 1319
		if (table_group_tmp->ops->create_table !=
				table_group->ops->create_table) {
1320 1321 1322 1323 1324 1325
			pr_warn("tce_vfio: Group %d is incompatible with group %d\n",
					iommu_group_id(iommu_group),
					iommu_group_id(tcegrp->grp));
			ret = -EPERM;
			goto unlock_exit;
		}
1326 1327
	}

1328 1329 1330
	tcegrp = kzalloc(sizeof(*tcegrp), GFP_KERNEL);
	if (!tcegrp) {
		ret = -ENOMEM;
1331 1332 1333
		goto unlock_exit;
	}

1334
	if (!table_group->ops || !table_group->ops->take_ownership ||
1335
			!table_group->ops->release_ownership) {
1336
		ret = tce_iommu_take_ownership(container, table_group);
1337
	} else {
1338
		ret = tce_iommu_take_ownership_ddw(container, table_group);
1339
		if (!tce_groups_attached(container) && !container->tables[0])
1340
			container->def_window_pending = true;
1341
	}
1342

1343 1344 1345 1346
	if (!ret) {
		tcegrp->grp = iommu_group;