1 // SPDX-License-Identifier: GPL-2.0-only 2 /* 3 * Copyright (c) 2006, Intel Corporation. 4 * 5 * Copyright (C) 2006-2008 Intel Corporation 6 * Author: Ashok Raj <ashok.raj@intel.com> 7 * Author: Shaohua Li <shaohua.li@intel.com> 8 * Author: Anil S Keshavamurthy <anil.s.keshavamurthy@intel.com> 9 * 10 * This file implements early detection/parsing of Remapping Devices 11 * reported to OS through BIOS via DMA remapping reporting (DMAR) ACPI 12 * tables. 13 * 14 * These routines are used by both DMA-remapping and Interrupt-remapping 15 */ 16 17 #define pr_fmt(fmt) "DMAR: " fmt 18 19 #include <linux/pci.h> 20 #include <linux/dmar.h> 21 #include <linux/iova.h> 22 #include <linux/intel-iommu.h> 23 #include <linux/timer.h> 24 #include <linux/irq.h> 25 #include <linux/interrupt.h> 26 #include <linux/tboot.h> 27 #include <linux/dmi.h> 28 #include <linux/slab.h> 29 #include <linux/iommu.h> 30 #include <linux/numa.h> 31 #include <linux/limits.h> 32 #include <asm/irq_remapping.h> 33 #include <asm/iommu_table.h> 34 #include <trace/events/intel_iommu.h> 35 36 #include "../irq_remapping.h" 37 #include "perf.h" 38 39 typedef int (*dmar_res_handler_t)(struct acpi_dmar_header *, void *); 40 struct dmar_res_callback { 41 dmar_res_handler_t cb[ACPI_DMAR_TYPE_RESERVED]; 42 void *arg[ACPI_DMAR_TYPE_RESERVED]; 43 bool ignore_unhandled; 44 bool print_entry; 45 }; 46 47 /* 48 * Assumptions: 49 * 1) The hotplug framework guarentees that DMAR unit will be hot-added 50 * before IO devices managed by that unit. 51 * 2) The hotplug framework guarantees that DMAR unit will be hot-removed 52 * after IO devices managed by that unit. 53 * 3) Hotplug events are rare. 54 * 55 * Locking rules for DMA and interrupt remapping related global data structures: 56 * 1) Use dmar_global_lock in process context 57 * 2) Use RCU in interrupt context 58 */ 59 DECLARE_RWSEM(dmar_global_lock); 60 LIST_HEAD(dmar_drhd_units); 61 62 struct acpi_table_header * __initdata dmar_tbl; 63 static int dmar_dev_scope_status = 1; 64 static unsigned long dmar_seq_ids[BITS_TO_LONGS(DMAR_UNITS_SUPPORTED)]; 65 66 static int alloc_iommu(struct dmar_drhd_unit *drhd); 67 static void free_iommu(struct intel_iommu *iommu); 68 69 extern const struct iommu_ops intel_iommu_ops; 70 71 static void dmar_register_drhd_unit(struct dmar_drhd_unit *drhd) 72 { 73 /* 74 * add INCLUDE_ALL at the tail, so scan the list will find it at 75 * the very end. 76 */ 77 if (drhd->include_all) 78 list_add_tail_rcu(&drhd->list, &dmar_drhd_units); 79 else 80 list_add_rcu(&drhd->list, &dmar_drhd_units); 81 } 82 83 void *dmar_alloc_dev_scope(void *start, void *end, int *cnt) 84 { 85 struct acpi_dmar_device_scope *scope; 86 87 *cnt = 0; 88 while (start < end) { 89 scope = start; 90 if (scope->entry_type == ACPI_DMAR_SCOPE_TYPE_NAMESPACE || 91 scope->entry_type == ACPI_DMAR_SCOPE_TYPE_ENDPOINT || 92 scope->entry_type == ACPI_DMAR_SCOPE_TYPE_BRIDGE) 93 (*cnt)++; 94 else if (scope->entry_type != ACPI_DMAR_SCOPE_TYPE_IOAPIC && 95 scope->entry_type != ACPI_DMAR_SCOPE_TYPE_HPET) { 96 pr_warn("Unsupported device scope\n"); 97 } 98 start += scope->length; 99 } 100 if (*cnt == 0) 101 return NULL; 102 103 return kcalloc(*cnt, sizeof(struct dmar_dev_scope), GFP_KERNEL); 104 } 105 106 void dmar_free_dev_scope(struct dmar_dev_scope **devices, int *cnt) 107 { 108 int i; 109 struct device *tmp_dev; 110 111 if (*devices && *cnt) { 112 for_each_active_dev_scope(*devices, *cnt, i, tmp_dev) 113 put_device(tmp_dev); 114 kfree(*devices); 115 } 116 117 *devices = NULL; 118 *cnt = 0; 119 } 120 121 /* Optimize out kzalloc()/kfree() for normal cases */ 122 static char dmar_pci_notify_info_buf[64]; 123 124 static struct dmar_pci_notify_info * 125 dmar_alloc_pci_notify_info(struct pci_dev *dev, unsigned long event) 126 { 127 int level = 0; 128 size_t size; 129 struct pci_dev *tmp; 130 struct dmar_pci_notify_info *info; 131 132 BUG_ON(dev->is_virtfn); 133 134 /* 135 * Ignore devices that have a domain number higher than what can 136 * be looked up in DMAR, e.g. VMD subdevices with domain 0x10000 137 */ 138 if (pci_domain_nr(dev->bus) > U16_MAX) 139 return NULL; 140 141 /* Only generate path[] for device addition event */ 142 if (event == BUS_NOTIFY_ADD_DEVICE) 143 for (tmp = dev; tmp; tmp = tmp->bus->self) 144 level++; 145 146 size = struct_size(info, path, level); 147 if (size <= sizeof(dmar_pci_notify_info_buf)) { 148 info = (struct dmar_pci_notify_info *)dmar_pci_notify_info_buf; 149 } else { 150 info = kzalloc(size, GFP_KERNEL); 151 if (!info) { 152 if (dmar_dev_scope_status == 0) 153 dmar_dev_scope_status = -ENOMEM; 154 return NULL; 155 } 156 } 157 158 info->event = event; 159 info->dev = dev; 160 info->seg = pci_domain_nr(dev->bus); 161 info->level = level; 162 if (event == BUS_NOTIFY_ADD_DEVICE) { 163 for (tmp = dev; tmp; tmp = tmp->bus->self) { 164 level--; 165 info->path[level].bus = tmp->bus->number; 166 info->path[level].device = PCI_SLOT(tmp->devfn); 167 info->path[level].function = PCI_FUNC(tmp->devfn); 168 if (pci_is_root_bus(tmp->bus)) 169 info->bus = tmp->bus->number; 170 } 171 } 172 173 return info; 174 } 175 176 static inline void dmar_free_pci_notify_info(struct dmar_pci_notify_info *info) 177 { 178 if ((void *)info != dmar_pci_notify_info_buf) 179 kfree(info); 180 } 181 182 static bool dmar_match_pci_path(struct dmar_pci_notify_info *info, int bus, 183 struct acpi_dmar_pci_path *path, int count) 184 { 185 int i; 186 187 if (info->bus != bus) 188 goto fallback; 189 if (info->level != count) 190 goto fallback; 191 192 for (i = 0; i < count; i++) { 193 if (path[i].device != info->path[i].device || 194 path[i].function != info->path[i].function) 195 goto fallback; 196 } 197 198 return true; 199 200 fallback: 201 202 if (count != 1) 203 return false; 204 205 i = info->level - 1; 206 if (bus == info->path[i].bus && 207 path[0].device == info->path[i].device && 208 path[0].function == info->path[i].function) { 209 pr_info(FW_BUG "RMRR entry for device %02x:%02x.%x is broken - applying workaround\n", 210 bus, path[0].device, path[0].function); 211 return true; 212 } 213 214 return false; 215 } 216 217 /* Return: > 0 if match found, 0 if no match found, < 0 if error happens */ 218 int dmar_insert_dev_scope(struct dmar_pci_notify_info *info, 219 void *start, void*end, u16 segment, 220 struct dmar_dev_scope *devices, 221 int devices_cnt) 222 { 223 int i, level; 224 struct device *tmp, *dev = &info->dev->dev; 225 struct acpi_dmar_device_scope *scope; 226 struct acpi_dmar_pci_path *path; 227 228 if (segment != info->seg) 229 return 0; 230 231 for (; start < end; start += scope->length) { 232 scope = start; 233 if (scope->entry_type != ACPI_DMAR_SCOPE_TYPE_ENDPOINT && 234 scope->entry_type != ACPI_DMAR_SCOPE_TYPE_BRIDGE) 235 continue; 236 237 path = (struct acpi_dmar_pci_path *)(scope + 1); 238 level = (scope->length - sizeof(*scope)) / sizeof(*path); 239 if (!dmar_match_pci_path(info, scope->bus, path, level)) 240 continue; 241 242 /* 243 * We expect devices with endpoint scope to have normal PCI 244 * headers, and devices with bridge scope to have bridge PCI 245 * headers. However PCI NTB devices may be listed in the 246 * DMAR table with bridge scope, even though they have a 247 * normal PCI header. NTB devices are identified by class 248 * "BRIDGE_OTHER" (0680h) - we don't declare a socpe mismatch 249 * for this special case. 250 */ 251 if ((scope->entry_type == ACPI_DMAR_SCOPE_TYPE_ENDPOINT && 252 info->dev->hdr_type != PCI_HEADER_TYPE_NORMAL) || 253 (scope->entry_type == ACPI_DMAR_SCOPE_TYPE_BRIDGE && 254 (info->dev->hdr_type == PCI_HEADER_TYPE_NORMAL && 255 info->dev->class >> 16 != PCI_BASE_CLASS_BRIDGE))) { 256 pr_warn("Device scope type does not match for %s\n", 257 pci_name(info->dev)); 258 return -EINVAL; 259 } 260 261 for_each_dev_scope(devices, devices_cnt, i, tmp) 262 if (tmp == NULL) { 263 devices[i].bus = info->dev->bus->number; 264 devices[i].devfn = info->dev->devfn; 265 rcu_assign_pointer(devices[i].dev, 266 get_device(dev)); 267 return 1; 268 } 269 BUG_ON(i >= devices_cnt); 270 } 271 272 return 0; 273 } 274 275 int dmar_remove_dev_scope(struct dmar_pci_notify_info *info, u16 segment, 276 struct dmar_dev_scope *devices, int count) 277 { 278 int index; 279 struct device *tmp; 280 281 if (info->seg != segment) 282 return 0; 283 284 for_each_active_dev_scope(devices, count, index, tmp) 285 if (tmp == &info->dev->dev) { 286 RCU_INIT_POINTER(devices[index].dev, NULL); 287 synchronize_rcu(); 288 put_device(tmp); 289 return 1; 290 } 291 292 return 0; 293 } 294 295 static int dmar_pci_bus_add_dev(struct dmar_pci_notify_info *info) 296 { 297 int ret = 0; 298 struct dmar_drhd_unit *dmaru; 299 struct acpi_dmar_hardware_unit *drhd; 300 301 for_each_drhd_unit(dmaru) { 302 if (dmaru->include_all) 303 continue; 304 305 drhd = container_of(dmaru->hdr, 306 struct acpi_dmar_hardware_unit, header); 307 ret = dmar_insert_dev_scope(info, (void *)(drhd + 1), 308 ((void *)drhd) + drhd->header.length, 309 dmaru->segment, 310 dmaru->devices, dmaru->devices_cnt); 311 if (ret) 312 break; 313 } 314 if (ret >= 0) 315 ret = dmar_iommu_notify_scope_dev(info); 316 if (ret < 0 && dmar_dev_scope_status == 0) 317 dmar_dev_scope_status = ret; 318 319 if (ret >= 0) 320 intel_irq_remap_add_device(info); 321 322 return ret; 323 } 324 325 static void dmar_pci_bus_del_dev(struct dmar_pci_notify_info *info) 326 { 327 struct dmar_drhd_unit *dmaru; 328 329 for_each_drhd_unit(dmaru) 330 if (dmar_remove_dev_scope(info, dmaru->segment, 331 dmaru->devices, dmaru->devices_cnt)) 332 break; 333 dmar_iommu_notify_scope_dev(info); 334 } 335 336 static inline void vf_inherit_msi_domain(struct pci_dev *pdev) 337 { 338 struct pci_dev *physfn = pci_physfn(pdev); 339 340 dev_set_msi_domain(&pdev->dev, dev_get_msi_domain(&physfn->dev)); 341 } 342 343 static int dmar_pci_bus_notifier(struct notifier_block *nb, 344 unsigned long action, void *data) 345 { 346 struct pci_dev *pdev = to_pci_dev(data); 347 struct dmar_pci_notify_info *info; 348 349 /* Only care about add/remove events for physical functions. 350 * For VFs we actually do the lookup based on the corresponding 351 * PF in device_to_iommu() anyway. */ 352 if (pdev->is_virtfn) { 353 /* 354 * Ensure that the VF device inherits the irq domain of the 355 * PF device. Ideally the device would inherit the domain 356 * from the bus, but DMAR can have multiple units per bus 357 * which makes this impossible. The VF 'bus' could inherit 358 * from the PF device, but that's yet another x86'sism to 359 * inflict on everybody else. 360 */ 361 if (action == BUS_NOTIFY_ADD_DEVICE) 362 vf_inherit_msi_domain(pdev); 363 return NOTIFY_DONE; 364 } 365 366 if (action != BUS_NOTIFY_ADD_DEVICE && 367 action != BUS_NOTIFY_REMOVED_DEVICE) 368 return NOTIFY_DONE; 369 370 info = dmar_alloc_pci_notify_info(pdev, action); 371 if (!info) 372 return NOTIFY_DONE; 373 374 down_write(&dmar_global_lock); 375 if (action == BUS_NOTIFY_ADD_DEVICE) 376 dmar_pci_bus_add_dev(info); 377 else if (action == BUS_NOTIFY_REMOVED_DEVICE) 378 dmar_pci_bus_del_dev(info); 379 up_write(&dmar_global_lock); 380 381 dmar_free_pci_notify_info(info); 382 383 return NOTIFY_OK; 384 } 385 386 static struct notifier_block dmar_pci_bus_nb = { 387 .notifier_call = dmar_pci_bus_notifier, 388 .priority = INT_MIN, 389 }; 390 391 static struct dmar_drhd_unit * 392 dmar_find_dmaru(struct acpi_dmar_hardware_unit *drhd) 393 { 394 struct dmar_drhd_unit *dmaru; 395 396 list_for_each_entry_rcu(dmaru, &dmar_drhd_units, list, 397 dmar_rcu_check()) 398 if (dmaru->segment == drhd->segment && 399 dmaru->reg_base_addr == drhd->address) 400 return dmaru; 401 402 return NULL; 403 } 404 405 /* 406 * dmar_parse_one_drhd - parses exactly one DMA remapping hardware definition 407 * structure which uniquely represent one DMA remapping hardware unit 408 * present in the platform 409 */ 410 static int dmar_parse_one_drhd(struct acpi_dmar_header *header, void *arg) 411 { 412 struct acpi_dmar_hardware_unit *drhd; 413 struct dmar_drhd_unit *dmaru; 414 int ret; 415 416 drhd = (struct acpi_dmar_hardware_unit *)header; 417 dmaru = dmar_find_dmaru(drhd); 418 if (dmaru) 419 goto out; 420 421 dmaru = kzalloc(sizeof(*dmaru) + header->length, GFP_KERNEL); 422 if (!dmaru) 423 return -ENOMEM; 424 425 /* 426 * If header is allocated from slab by ACPI _DSM method, we need to 427 * copy the content because the memory buffer will be freed on return. 428 */ 429 dmaru->hdr = (void *)(dmaru + 1); 430 memcpy(dmaru->hdr, header, header->length); 431 dmaru->reg_base_addr = drhd->address; 432 dmaru->segment = drhd->segment; 433 dmaru->include_all = drhd->flags & 0x1; /* BIT0: INCLUDE_ALL */ 434 dmaru->devices = dmar_alloc_dev_scope((void *)(drhd + 1), 435 ((void *)drhd) + drhd->header.length, 436 &dmaru->devices_cnt); 437 if (dmaru->devices_cnt && dmaru->devices == NULL) { 438 kfree(dmaru); 439 return -ENOMEM; 440 } 441 442 ret = alloc_iommu(dmaru); 443 if (ret) { 444 dmar_free_dev_scope(&dmaru->devices, 445 &dmaru->devices_cnt); 446 kfree(dmaru); 447 return ret; 448 } 449 dmar_register_drhd_unit(dmaru); 450 451 out: 452 if (arg) 453 (*(int *)arg)++; 454 455 return 0; 456 } 457 458 static void dmar_free_drhd(struct dmar_drhd_unit *dmaru) 459 { 460 if (dmaru->devices && dmaru->devices_cnt) 461 dmar_free_dev_scope(&dmaru->devices, &dmaru->devices_cnt); 462 if (dmaru->iommu) 463 free_iommu(dmaru->iommu); 464 kfree(dmaru); 465 } 466 467 static int __init dmar_parse_one_andd(struct acpi_dmar_header *header, 468 void *arg) 469 { 470 struct acpi_dmar_andd *andd = (void *)header; 471 472 /* Check for NUL termination within the designated length */ 473 if (strnlen(andd->device_name, header->length - 8) == header->length - 8) { 474 pr_warn(FW_BUG 475 "Your BIOS is broken; ANDD object name is not NUL-terminated\n" 476 "BIOS vendor: %s; Ver: %s; Product Version: %s\n", 477 dmi_get_system_info(DMI_BIOS_VENDOR), 478 dmi_get_system_info(DMI_BIOS_VERSION), 479 dmi_get_system_info(DMI_PRODUCT_VERSION)); 480 add_taint(TAINT_FIRMWARE_WORKAROUND, LOCKDEP_STILL_OK); 481 return -EINVAL; 482 } 483 pr_info("ANDD device: %x name: %s\n", andd->device_number, 484 andd->device_name); 485 486 return 0; 487 } 488 489 #ifdef CONFIG_ACPI_NUMA 490 static int dmar_parse_one_rhsa(struct acpi_dmar_header *header, void *arg) 491 { 492 struct acpi_dmar_rhsa *rhsa; 493 struct dmar_drhd_unit *drhd; 494 495 rhsa = (struct acpi_dmar_rhsa *)header; 496 for_each_drhd_unit(drhd) { 497 if (drhd->reg_base_addr == rhsa->base_address) { 498 int node = pxm_to_node(rhsa->proximity_domain); 499 500 if (!node_online(node)) 501 node = NUMA_NO_NODE; 502 drhd->iommu->node = node; 503 return 0; 504 } 505 } 506 pr_warn(FW_BUG 507 "Your BIOS is broken; RHSA refers to non-existent DMAR unit at %llx\n" 508 "BIOS vendor: %s; Ver: %s; Product Version: %s\n", 509 rhsa->base_address, 510 dmi_get_system_info(DMI_BIOS_VENDOR), 511 dmi_get_system_info(DMI_BIOS_VERSION), 512 dmi_get_system_info(DMI_PRODUCT_VERSION)); 513 add_taint(TAINT_FIRMWARE_WORKAROUND, LOCKDEP_STILL_OK); 514 515 return 0; 516 } 517 #else 518 #define dmar_parse_one_rhsa dmar_res_noop 519 #endif 520 521 static void 522 dmar_table_print_dmar_entry(struct acpi_dmar_header *header) 523 { 524 struct acpi_dmar_hardware_unit *drhd; 525 struct acpi_dmar_reserved_memory *rmrr; 526 struct acpi_dmar_atsr *atsr; 527 struct acpi_dmar_rhsa *rhsa; 528 struct acpi_dmar_satc *satc; 529 530 switch (header->type) { 531 case ACPI_DMAR_TYPE_HARDWARE_UNIT: 532 drhd = container_of(header, struct acpi_dmar_hardware_unit, 533 header); 534 pr_info("DRHD base: %#016Lx flags: %#x\n", 535 (unsigned long long)drhd->address, drhd->flags); 536 break; 537 case ACPI_DMAR_TYPE_RESERVED_MEMORY: 538 rmrr = container_of(header, struct acpi_dmar_reserved_memory, 539 header); 540 pr_info("RMRR base: %#016Lx end: %#016Lx\n", 541 (unsigned long long)rmrr->base_address, 542 (unsigned long long)rmrr->end_address); 543 break; 544 case ACPI_DMAR_TYPE_ROOT_ATS: 545 atsr = container_of(header, struct acpi_dmar_atsr, header); 546 pr_info("ATSR flags: %#x\n", atsr->flags); 547 break; 548 case ACPI_DMAR_TYPE_HARDWARE_AFFINITY: 549 rhsa = container_of(header, struct acpi_dmar_rhsa, header); 550 pr_info("RHSA base: %#016Lx proximity domain: %#x\n", 551 (unsigned long long)rhsa->base_address, 552 rhsa->proximity_domain); 553 break; 554 case ACPI_DMAR_TYPE_NAMESPACE: 555 /* We don't print this here because we need to sanity-check 556 it first. So print it in dmar_parse_one_andd() instead. */ 557 break; 558 case ACPI_DMAR_TYPE_SATC: 559 satc = container_of(header, struct acpi_dmar_satc, header); 560 pr_info("SATC flags: 0x%x\n", satc->flags); 561 break; 562 } 563 } 564 565 /** 566 * dmar_table_detect - checks to see if the platform supports DMAR devices 567 */ 568 static int __init dmar_table_detect(void) 569 { 570 acpi_status status = AE_OK; 571 572 /* if we could find DMAR table, then there are DMAR devices */ 573 status = acpi_get_table(ACPI_SIG_DMAR, 0, &dmar_tbl); 574 575 if (ACPI_SUCCESS(status) && !dmar_tbl) { 576 pr_warn("Unable to map DMAR\n"); 577 status = AE_NOT_FOUND; 578 } 579 580 return ACPI_SUCCESS(status) ? 0 : -ENOENT; 581 } 582 583 static int dmar_walk_remapping_entries(struct acpi_dmar_header *start, 584 size_t len, struct dmar_res_callback *cb) 585 { 586 struct acpi_dmar_header *iter, *next; 587 struct acpi_dmar_header *end = ((void *)start) + len; 588 589 for (iter = start; iter < end; iter = next) { 590 next = (void *)iter + iter->length; 591 if (iter->length == 0) { 592 /* Avoid looping forever on bad ACPI tables */ 593 pr_debug(FW_BUG "Invalid 0-length structure\n"); 594 break; 595 } else if (next > end) { 596 /* Avoid passing table end */ 597 pr_warn(FW_BUG "Record passes table end\n"); 598 return -EINVAL; 599 } 600 601 if (cb->print_entry) 602 dmar_table_print_dmar_entry(iter); 603 604 if (iter->type >= ACPI_DMAR_TYPE_RESERVED) { 605 /* continue for forward compatibility */ 606 pr_debug("Unknown DMAR structure type %d\n", 607 iter->type); 608 } else if (cb->cb[iter->type]) { 609 int ret; 610 611 ret = cb->cb[iter->type](iter, cb->arg[iter->type]); 612 if (ret) 613 return ret; 614 } else if (!cb->ignore_unhandled) { 615 pr_warn("No handler for DMAR structure type %d\n", 616 iter->type); 617 return -EINVAL; 618 } 619 } 620 621 return 0; 622 } 623 624 static inline int dmar_walk_dmar_table(struct acpi_table_dmar *dmar, 625 struct dmar_res_callback *cb) 626 { 627 return dmar_walk_remapping_entries((void *)(dmar + 1), 628 dmar->header.length - sizeof(*dmar), cb); 629 } 630 631 /** 632 * parse_dmar_table - parses the DMA reporting table 633 */ 634 static int __init 635 parse_dmar_table(void) 636 { 637 struct acpi_table_dmar *dmar; 638 int drhd_count = 0; 639 int ret; 640 struct dmar_res_callback cb = { 641 .print_entry = true, 642 .ignore_unhandled = true, 643 .arg[ACPI_DMAR_TYPE_HARDWARE_UNIT] = &drhd_count, 644 .cb[ACPI_DMAR_TYPE_HARDWARE_UNIT] = &dmar_parse_one_drhd, 645 .cb[ACPI_DMAR_TYPE_RESERVED_MEMORY] = &dmar_parse_one_rmrr, 646 .cb[ACPI_DMAR_TYPE_ROOT_ATS] = &dmar_parse_one_atsr, 647 .cb[ACPI_DMAR_TYPE_HARDWARE_AFFINITY] = &dmar_parse_one_rhsa, 648 .cb[ACPI_DMAR_TYPE_NAMESPACE] = &dmar_parse_one_andd, 649 .cb[ACPI_DMAR_TYPE_SATC] = &dmar_parse_one_satc, 650 }; 651 652 /* 653 * Do it again, earlier dmar_tbl mapping could be mapped with 654 * fixed map. 655 */ 656 dmar_table_detect(); 657 658 /* 659 * ACPI tables may not be DMA protected by tboot, so use DMAR copy 660 * SINIT saved in SinitMleData in TXT heap (which is DMA protected) 661 */ 662 dmar_tbl = tboot_get_dmar_table(dmar_tbl); 663 664 dmar = (struct acpi_table_dmar *)dmar_tbl; 665 if (!dmar) 666 return -ENODEV; 667 668 if (dmar->width < PAGE_SHIFT - 1) { 669 pr_warn("Invalid DMAR haw\n"); 670 return -EINVAL; 671 } 672 673 pr_info("Host address width %d\n", dmar->width + 1); 674 ret = dmar_walk_dmar_table(dmar, &cb); 675 if (ret == 0 && drhd_count == 0) 676 pr_warn(FW_BUG "No DRHD structure found in DMAR table\n"); 677 678 return ret; 679 } 680 681 static int dmar_pci_device_match(struct dmar_dev_scope devices[], 682 int cnt, struct pci_dev *dev) 683 { 684 int index; 685 struct device *tmp; 686 687 while (dev) { 688 for_each_active_dev_scope(devices, cnt, index, tmp) 689 if (dev_is_pci(tmp) && dev == to_pci_dev(tmp)) 690 return 1; 691 692 /* Check our parent */ 693 dev = dev->bus->self; 694 } 695 696 return 0; 697 } 698 699 struct dmar_drhd_unit * 700 dmar_find_matched_drhd_unit(struct pci_dev *dev) 701 { 702 struct dmar_drhd_unit *dmaru; 703 struct acpi_dmar_hardware_unit *drhd; 704 705 dev = pci_physfn(dev); 706 707 rcu_read_lock(); 708 for_each_drhd_unit(dmaru) { 709 drhd = container_of(dmaru->hdr, 710 struct acpi_dmar_hardware_unit, 711 header); 712 713 if (dmaru->include_all && 714 drhd->segment == pci_domain_nr(dev->bus)) 715 goto out; 716 717 if (dmar_pci_device_match(dmaru->devices, 718 dmaru->devices_cnt, dev)) 719 goto out; 720 } 721 dmaru = NULL; 722 out: 723 rcu_read_unlock(); 724 725 return dmaru; 726 } 727 728 static void __init dmar_acpi_insert_dev_scope(u8 device_number, 729 struct acpi_device *adev) 730 { 731 struct dmar_drhd_unit *dmaru; 732 struct acpi_dmar_hardware_unit *drhd; 733 struct acpi_dmar_device_scope *scope; 734 struct device *tmp; 735 int i; 736 struct acpi_dmar_pci_path *path; 737 738 for_each_drhd_unit(dmaru) { 739 drhd = container_of(dmaru->hdr, 740 struct acpi_dmar_hardware_unit, 741 header); 742 743 for (scope = (void *)(drhd + 1); 744 (unsigned long)scope < ((unsigned long)drhd) + drhd->header.length; 745 scope = ((void *)scope) + scope->length) { 746 if (scope->entry_type != ACPI_DMAR_SCOPE_TYPE_NAMESPACE) 747 continue; 748 if (scope->enumeration_id != device_number) 749 continue; 750 751 path = (void *)(scope + 1); 752 pr_info("ACPI device \"%s\" under DMAR at %llx as %02x:%02x.%d\n", 753 dev_name(&adev->dev), dmaru->reg_base_addr, 754 scope->bus, path->device, path->function); 755 for_each_dev_scope(dmaru->devices, dmaru->devices_cnt, i, tmp) 756 if (tmp == NULL) { 757 dmaru->devices[i].bus = scope->bus; 758 dmaru->devices[i].devfn = PCI_DEVFN(path->device, 759 path->function); 760 rcu_assign_pointer(dmaru->devices[i].dev, 761 get_device(&adev->dev)); 762 return; 763 } 764 BUG_ON(i >= dmaru->devices_cnt); 765 } 766 } 767 pr_warn("No IOMMU scope found for ANDD enumeration ID %d (%s)\n", 768 device_number, dev_name(&adev->dev)); 769 } 770 771 static int __init dmar_acpi_dev_scope_init(void) 772 { 773 struct acpi_dmar_andd *andd; 774 775 if (dmar_tbl == NULL) 776 return -ENODEV; 777 778 for (andd = (void *)dmar_tbl + sizeof(struct acpi_table_dmar); 779 ((unsigned long)andd) < ((unsigned long)dmar_tbl) + dmar_tbl->length; 780 andd = ((void *)andd) + andd->header.length) { 781 if (andd->header.type == ACPI_DMAR_TYPE_NAMESPACE) { 782 acpi_handle h; 783 struct acpi_device *adev; 784 785 if (!ACPI_SUCCESS(acpi_get_handle(ACPI_ROOT_OBJECT, 786 andd->device_name, 787 &h))) { 788 pr_err("Failed to find handle for ACPI object %s\n", 789 andd->device_name); 790 continue; 791 } 792 if (acpi_bus_get_device(h, &adev)) { 793 pr_err("Failed to get device for ACPI object %s\n", 794 andd->device_name); 795 continue; 796 } 797 dmar_acpi_insert_dev_scope(andd->device_number, adev); 798 } 799 } 800 return 0; 801 } 802 803 int __init dmar_dev_scope_init(void) 804 { 805 struct pci_dev *dev = NULL; 806 struct dmar_pci_notify_info *info; 807 808 if (dmar_dev_scope_status != 1) 809 return dmar_dev_scope_status; 810 811 if (list_empty(&dmar_drhd_units)) { 812 dmar_dev_scope_status = -ENODEV; 813 } else { 814 dmar_dev_scope_status = 0; 815 816 dmar_acpi_dev_scope_init(); 817 818 for_each_pci_dev(dev) { 819 if (dev->is_virtfn) 820 continue; 821 822 info = dmar_alloc_pci_notify_info(dev, 823 BUS_NOTIFY_ADD_DEVICE); 824 if (!info) { 825 return dmar_dev_scope_status; 826 } else { 827 dmar_pci_bus_add_dev(info); 828 dmar_free_pci_notify_info(info); 829 } 830 } 831 } 832 833 return dmar_dev_scope_status; 834 } 835 836 void __init dmar_register_bus_notifier(void) 837 { 838 bus_register_notifier(&pci_bus_type, &dmar_pci_bus_nb); 839 } 840 841 842 int __init dmar_table_init(void) 843 { 844 static int dmar_table_initialized; 845 int ret; 846 847 if (dmar_table_initialized == 0) { 848 ret = parse_dmar_table(); 849 if (ret < 0) { 850 if (ret != -ENODEV) 851 pr_info("Parse DMAR table failure.\n"); 852 } else if (list_empty(&dmar_drhd_units)) { 853 pr_info("No DMAR devices found\n"); 854 ret = -ENODEV; 855 } 856 857 if (ret < 0) 858 dmar_table_initialized = ret; 859 else 860 dmar_table_initialized = 1; 861 } 862 863 return dmar_table_initialized < 0 ? dmar_table_initialized : 0; 864 } 865 866 static void warn_invalid_dmar(u64 addr, const char *message) 867 { 868 pr_warn_once(FW_BUG 869 "Your BIOS is broken; DMAR reported at address %llx%s!\n" 870 "BIOS vendor: %s; Ver: %s; Product Version: %s\n", 871 addr, message, 872 dmi_get_system_info(DMI_BIOS_VENDOR), 873 dmi_get_system_info(DMI_BIOS_VERSION), 874 dmi_get_system_info(DMI_PRODUCT_VERSION)); 875 add_taint(TAINT_FIRMWARE_WORKAROUND, LOCKDEP_STILL_OK); 876 } 877 878 static int __ref 879 dmar_validate_one_drhd(struct acpi_dmar_header *entry, void *arg) 880 { 881 struct acpi_dmar_hardware_unit *drhd; 882 void __iomem *addr; 883 u64 cap, ecap; 884 885 drhd = (void *)entry; 886 if (!drhd->address) { 887 warn_invalid_dmar(0, ""); 888 return -EINVAL; 889 } 890 891 if (arg) 892 addr = ioremap(drhd->address, VTD_PAGE_SIZE); 893 else 894 addr = early_ioremap(drhd->address, VTD_PAGE_SIZE); 895 if (!addr) { 896 pr_warn("Can't validate DRHD address: %llx\n", drhd->address); 897 return -EINVAL; 898 } 899 900 cap = dmar_readq(addr + DMAR_CAP_REG); 901 ecap = dmar_readq(addr + DMAR_ECAP_REG); 902 903 if (arg) 904 iounmap(addr); 905 else 906 early_iounmap(addr, VTD_PAGE_SIZE); 907 908 if (cap == (uint64_t)-1 && ecap == (uint64_t)-1) { 909 warn_invalid_dmar(drhd->address, " returns all ones"); 910 return -EINVAL; 911 } 912 913 return 0; 914 } 915 916 int __init detect_intel_iommu(void) 917 { 918 int ret; 919 struct dmar_res_callback validate_drhd_cb = { 920 .cb[ACPI_DMAR_TYPE_HARDWARE_UNIT] = &dmar_validate_one_drhd, 921 .ignore_unhandled = true, 922 }; 923 924 down_write(&dmar_global_lock); 925 ret = dmar_table_detect(); 926 if (!ret) 927 ret = dmar_walk_dmar_table((struct acpi_table_dmar *)dmar_tbl, 928 &validate_drhd_cb); 929 if (!ret && !no_iommu && !iommu_detected && 930 (!dmar_disabled || dmar_platform_optin())) { 931 iommu_detected = 1; 932 /* Make sure ACS will be enabled */ 933 pci_request_acs(); 934 } 935 936 #ifdef CONFIG_X86 937 if (!ret) { 938 x86_init.iommu.iommu_init = intel_iommu_init; 939 x86_platform.iommu_shutdown = intel_iommu_shutdown; 940 } 941 942 #endif 943 944 if (dmar_tbl) { 945 acpi_put_table(dmar_tbl); 946 dmar_tbl = NULL; 947 } 948 up_write(&dmar_global_lock); 949 950 return ret ? ret : 1; 951 } 952 953 static void unmap_iommu(struct intel_iommu *iommu) 954 { 955 iounmap(iommu->reg); 956 release_mem_region(iommu->reg_phys, iommu->reg_size); 957 } 958 959 /** 960 * map_iommu: map the iommu's registers 961 * @iommu: the iommu to map 962 * @phys_addr: the physical address of the base resgister 963 * 964 * Memory map the iommu's registers. Start w/ a single page, and 965 * possibly expand if that turns out to be insufficent. 966 */ 967 static int map_iommu(struct intel_iommu *iommu, u64 phys_addr) 968 { 969 int map_size, err=0; 970 971 iommu->reg_phys = phys_addr; 972 iommu->reg_size = VTD_PAGE_SIZE; 973 974 if (!request_mem_region(iommu->reg_phys, iommu->reg_size, iommu->name)) { 975 pr_err("Can't reserve memory\n"); 976 err = -EBUSY; 977 goto out; 978 } 979 980 iommu->reg = ioremap(iommu->reg_phys, iommu->reg_size); 981 if (!iommu->reg) { 982 pr_err("Can't map the region\n"); 983 err = -ENOMEM; 984 goto release; 985 } 986 987 iommu->cap = dmar_readq(iommu->reg + DMAR_CAP_REG); 988 iommu->ecap = dmar_readq(iommu->reg + DMAR_ECAP_REG); 989 990 if (iommu->cap == (uint64_t)-1 && iommu->ecap == (uint64_t)-1) { 991 err = -EINVAL; 992 warn_invalid_dmar(phys_addr, " returns all ones"); 993 goto unmap; 994 } 995 if (ecap_vcs(iommu->ecap)) 996 iommu->vccap = dmar_readq(iommu->reg + DMAR_VCCAP_REG); 997 998 /* the registers might be more than one page */ 999 map_size = max_t(int, ecap_max_iotlb_offset(iommu->ecap), 1000 cap_max_fault_reg_offset(iommu->cap)); 1001 map_size = VTD_PAGE_ALIGN(map_size); 1002 if (map_size > iommu->reg_size) { 1003 iounmap(iommu->reg); 1004 release_mem_region(iommu->reg_phys, iommu->reg_size); 1005 iommu->reg_size = map_size; 1006 if (!request_mem_region(iommu->reg_phys, iommu->reg_size, 1007 iommu->name)) { 1008 pr_err("Can't reserve memory\n"); 1009 err = -EBUSY; 1010 goto out; 1011 } 1012 iommu->reg = ioremap(iommu->reg_phys, iommu->reg_size); 1013 if (!iommu->reg) { 1014 pr_err("Can't map the region\n"); 1015 err = -ENOMEM; 1016 goto release; 1017 } 1018 } 1019 err = 0; 1020 goto out; 1021 1022 unmap: 1023 iounmap(iommu->reg); 1024 release: 1025 release_mem_region(iommu->reg_phys, iommu->reg_size); 1026 out: 1027 return err; 1028 } 1029 1030 static int dmar_alloc_seq_id(struct intel_iommu *iommu) 1031 { 1032 iommu->seq_id = find_first_zero_bit(dmar_seq_ids, 1033 DMAR_UNITS_SUPPORTED); 1034 if (iommu->seq_id >= DMAR_UNITS_SUPPORTED) { 1035 iommu->seq_id = -1; 1036 } else { 1037 set_bit(iommu->seq_id, dmar_seq_ids); 1038 sprintf(iommu->name, "dmar%d", iommu->seq_id); 1039 } 1040 1041 return iommu->seq_id; 1042 } 1043 1044 static void dmar_free_seq_id(struct intel_iommu *iommu) 1045 { 1046 if (iommu->seq_id >= 0) { 1047 clear_bit(iommu->seq_id, dmar_seq_ids); 1048 iommu->seq_id = -1; 1049 } 1050 } 1051 1052 static int alloc_iommu(struct dmar_drhd_unit *drhd) 1053 { 1054 struct intel_iommu *iommu; 1055 u32 ver, sts; 1056 int agaw = -1; 1057 int msagaw = -1; 1058 int err; 1059 1060 if (!drhd->reg_base_addr) { 1061 warn_invalid_dmar(0, ""); 1062 return -EINVAL; 1063 } 1064 1065 iommu = kzalloc(sizeof(*iommu), GFP_KERNEL); 1066 if (!iommu) 1067 return -ENOMEM; 1068 1069 if (dmar_alloc_seq_id(iommu) < 0) { 1070 pr_err("Failed to allocate seq_id\n"); 1071 err = -ENOSPC; 1072 goto error; 1073 } 1074 1075 err = map_iommu(iommu, drhd->reg_base_addr); 1076 if (err) { 1077 pr_err("Failed to map %s\n", iommu->name); 1078 goto error_free_seq_id; 1079 } 1080 1081 err = -EINVAL; 1082 if (cap_sagaw(iommu->cap) == 0) { 1083 pr_info("%s: No supported address widths. Not attempting DMA translation.\n", 1084 iommu->name); 1085 drhd->ignored = 1; 1086 } 1087 1088 if (!drhd->ignored) { 1089 agaw = iommu_calculate_agaw(iommu); 1090 if (agaw < 0) { 1091 pr_err("Cannot get a valid agaw for iommu (seq_id = %d)\n", 1092 iommu->seq_id); 1093 drhd->ignored = 1; 1094 } 1095 } 1096 if (!drhd->ignored) { 1097 msagaw = iommu_calculate_max_sagaw(iommu); 1098 if (msagaw < 0) { 1099 pr_err("Cannot get a valid max agaw for iommu (seq_id = %d)\n", 1100 iommu->seq_id); 1101 drhd->ignored = 1; 1102 agaw = -1; 1103 } 1104 } 1105 iommu->agaw = agaw; 1106 iommu->msagaw = msagaw; 1107 iommu->segment = drhd->segment; 1108 1109 iommu->node = NUMA_NO_NODE; 1110 1111 ver = readl(iommu->reg + DMAR_VER_REG); 1112 pr_info("%s: reg_base_addr %llx ver %d:%d cap %llx ecap %llx\n", 1113 iommu->name, 1114 (unsigned long long)drhd->reg_base_addr, 1115 DMAR_VER_MAJOR(ver), DMAR_VER_MINOR(ver), 1116 (unsigned long long)iommu->cap, 1117 (unsigned long long)iommu->ecap); 1118 1119 /* Reflect status in gcmd */ 1120 sts = readl(iommu->reg + DMAR_GSTS_REG); 1121 if (sts & DMA_GSTS_IRES) 1122 iommu->gcmd |= DMA_GCMD_IRE; 1123 if (sts & DMA_GSTS_TES) 1124 iommu->gcmd |= DMA_GCMD_TE; 1125 if (sts & DMA_GSTS_QIES) 1126 iommu->gcmd |= DMA_GCMD_QIE; 1127 1128 raw_spin_lock_init(&iommu->register_lock); 1129 1130 /* 1131 * This is only for hotplug; at boot time intel_iommu_enabled won't 1132 * be set yet. When intel_iommu_init() runs, it registers the units 1133 * present at boot time, then sets intel_iommu_enabled. 1134 */ 1135 if (intel_iommu_enabled && !drhd->ignored) { 1136 err = iommu_device_sysfs_add(&iommu->iommu, NULL, 1137 intel_iommu_groups, 1138 "%s", iommu->name); 1139 if (err) 1140 goto err_unmap; 1141 1142 err = iommu_device_register(&iommu->iommu, &intel_iommu_ops, NULL); 1143 if (err) 1144 goto err_sysfs; 1145 } 1146 1147 drhd->iommu = iommu; 1148 iommu->drhd = drhd; 1149 1150 return 0; 1151 1152 err_sysfs: 1153 iommu_device_sysfs_remove(&iommu->iommu); 1154 err_unmap: 1155 unmap_iommu(iommu); 1156 error_free_seq_id: 1157 dmar_free_seq_id(iommu); 1158 error: 1159 kfree(iommu); 1160 return err; 1161 } 1162 1163 static void free_iommu(struct intel_iommu *iommu) 1164 { 1165 if (intel_iommu_enabled && !iommu->drhd->ignored) { 1166 iommu_device_unregister(&iommu->iommu); 1167 iommu_device_sysfs_remove(&iommu->iommu); 1168 } 1169 1170 if (iommu->irq) { 1171 if (iommu->pr_irq) { 1172 free_irq(iommu->pr_irq, iommu); 1173 dmar_free_hwirq(iommu->pr_irq); 1174 iommu->pr_irq = 0; 1175 } 1176 free_irq(iommu->irq, iommu); 1177 dmar_free_hwirq(iommu->irq); 1178 iommu->irq = 0; 1179 } 1180 1181 if (iommu->qi) { 1182 free_page((unsigned long)iommu->qi->desc); 1183 kfree(iommu->qi->desc_status); 1184 kfree(iommu->qi); 1185 } 1186 1187 if (iommu->reg) 1188 unmap_iommu(iommu); 1189 1190 dmar_free_seq_id(iommu); 1191 kfree(iommu); 1192 } 1193 1194 /* 1195 * Reclaim all the submitted descriptors which have completed its work. 1196 */ 1197 static inline void reclaim_free_desc(struct q_inval *qi) 1198 { 1199 while (qi->desc_status[qi->free_tail] == QI_DONE || 1200 qi->desc_status[qi->free_tail] == QI_ABORT) { 1201 qi->desc_status[qi->free_tail] = QI_FREE; 1202 qi->free_tail = (qi->free_tail + 1) % QI_LENGTH; 1203 qi->free_cnt++; 1204 } 1205 } 1206 1207 static const char *qi_type_string(u8 type) 1208 { 1209 switch (type) { 1210 case QI_CC_TYPE: 1211 return "Context-cache Invalidation"; 1212 case QI_IOTLB_TYPE: 1213 return "IOTLB Invalidation"; 1214 case QI_DIOTLB_TYPE: 1215 return "Device-TLB Invalidation"; 1216 case QI_IEC_TYPE: 1217 return "Interrupt Entry Cache Invalidation"; 1218 case QI_IWD_TYPE: 1219 return "Invalidation Wait"; 1220 case QI_EIOTLB_TYPE: 1221 return "PASID-based IOTLB Invalidation"; 1222 case QI_PC_TYPE: 1223 return "PASID-cache Invalidation"; 1224 case QI_DEIOTLB_TYPE: 1225 return "PASID-based Device-TLB Invalidation"; 1226 case QI_PGRP_RESP_TYPE: 1227 return "Page Group Response"; 1228 default: 1229 return "UNKNOWN"; 1230 } 1231 } 1232 1233 static void qi_dump_fault(struct intel_iommu *iommu, u32 fault) 1234 { 1235 unsigned int head = dmar_readl(iommu->reg + DMAR_IQH_REG); 1236 u64 iqe_err = dmar_readq(iommu->reg + DMAR_IQER_REG); 1237 struct qi_desc *desc = iommu->qi->desc + head; 1238 1239 if (fault & DMA_FSTS_IQE) 1240 pr_err("VT-d detected Invalidation Queue Error: Reason %llx", 1241 DMAR_IQER_REG_IQEI(iqe_err)); 1242 if (fault & DMA_FSTS_ITE) 1243 pr_err("VT-d detected Invalidation Time-out Error: SID %llx", 1244 DMAR_IQER_REG_ITESID(iqe_err)); 1245 if (fault & DMA_FSTS_ICE) 1246 pr_err("VT-d detected Invalidation Completion Error: SID %llx", 1247 DMAR_IQER_REG_ICESID(iqe_err)); 1248 1249 pr_err("QI HEAD: %s qw0 = 0x%llx, qw1 = 0x%llx\n", 1250 qi_type_string(desc->qw0 & 0xf), 1251 (unsigned long long)desc->qw0, 1252 (unsigned long long)desc->qw1); 1253 1254 head = ((head >> qi_shift(iommu)) + QI_LENGTH - 1) % QI_LENGTH; 1255 head <<= qi_shift(iommu); 1256 desc = iommu->qi->desc + head; 1257 1258 pr_err("QI PRIOR: %s qw0 = 0x%llx, qw1 = 0x%llx\n", 1259 qi_type_string(desc->qw0 & 0xf), 1260 (unsigned long long)desc->qw0, 1261 (unsigned long long)desc->qw1); 1262 } 1263 1264 static int qi_check_fault(struct intel_iommu *iommu, int index, int wait_index) 1265 { 1266 u32 fault; 1267 int head, tail; 1268 struct q_inval *qi = iommu->qi; 1269 int shift = qi_shift(iommu); 1270 1271 if (qi->desc_status[wait_index] == QI_ABORT) 1272 return -EAGAIN; 1273 1274 fault = readl(iommu->reg + DMAR_FSTS_REG); 1275 if (fault & (DMA_FSTS_IQE | DMA_FSTS_ITE | DMA_FSTS_ICE)) 1276 qi_dump_fault(iommu, fault); 1277 1278 /* 1279 * If IQE happens, the head points to the descriptor associated 1280 * with the error. No new descriptors are fetched until the IQE 1281 * is cleared. 1282 */ 1283 if (fault & DMA_FSTS_IQE) { 1284 head = readl(iommu->reg + DMAR_IQH_REG); 1285 if ((head >> shift) == index) { 1286 struct qi_desc *desc = qi->desc + head; 1287 1288 /* 1289 * desc->qw2 and desc->qw3 are either reserved or 1290 * used by software as private data. We won't print 1291 * out these two qw's for security consideration. 1292 */ 1293 memcpy(desc, qi->desc + (wait_index << shift), 1294 1 << shift); 1295 writel(DMA_FSTS_IQE, iommu->reg + DMAR_FSTS_REG); 1296 pr_info("Invalidation Queue Error (IQE) cleared\n"); 1297 return -EINVAL; 1298 } 1299 } 1300 1301 /* 1302 * If ITE happens, all pending wait_desc commands are aborted. 1303 * No new descriptors are fetched until the ITE is cleared. 1304 */ 1305 if (fault & DMA_FSTS_ITE) { 1306 head = readl(iommu->reg + DMAR_IQH_REG); 1307 head = ((head >> shift) - 1 + QI_LENGTH) % QI_LENGTH; 1308 head |= 1; 1309 tail = readl(iommu->reg + DMAR_IQT_REG); 1310 tail = ((tail >> shift) - 1 + QI_LENGTH) % QI_LENGTH; 1311 1312 writel(DMA_FSTS_ITE, iommu->reg + DMAR_FSTS_REG); 1313 pr_info("Invalidation Time-out Error (ITE) cleared\n"); 1314 1315 do { 1316 if (qi->desc_status[head] == QI_IN_USE) 1317 qi->desc_status[head] = QI_ABORT; 1318 head = (head - 2 + QI_LENGTH) % QI_LENGTH; 1319 } while (head != tail); 1320 1321 if (qi->desc_status[wait_index] == QI_ABORT) 1322 return -EAGAIN; 1323 } 1324 1325 if (fault & DMA_FSTS_ICE) { 1326 writel(DMA_FSTS_ICE, iommu->reg + DMAR_FSTS_REG); 1327 pr_info("Invalidation Completion Error (ICE) cleared\n"); 1328 } 1329 1330 return 0; 1331 } 1332 1333 /* 1334 * Function to submit invalidation descriptors of all types to the queued 1335 * invalidation interface(QI). Multiple descriptors can be submitted at a 1336 * time, a wait descriptor will be appended to each submission to ensure 1337 * hardware has completed the invalidation before return. Wait descriptors 1338 * can be part of the submission but it will not be polled for completion. 1339 */ 1340 int qi_submit_sync(struct intel_iommu *iommu, struct qi_desc *desc, 1341 unsigned int count, unsigned long options) 1342 { 1343 struct q_inval *qi = iommu->qi; 1344 s64 devtlb_start_ktime = 0; 1345 s64 iotlb_start_ktime = 0; 1346 s64 iec_start_ktime = 0; 1347 struct qi_desc wait_desc; 1348 int wait_index, index; 1349 unsigned long flags; 1350 int offset, shift; 1351 int rc, i; 1352 u64 type; 1353 1354 if (!qi) 1355 return 0; 1356 1357 type = desc->qw0 & GENMASK_ULL(3, 0); 1358 1359 if ((type == QI_IOTLB_TYPE || type == QI_EIOTLB_TYPE) && 1360 dmar_latency_enabled(iommu, DMAR_LATENCY_INV_IOTLB)) 1361 iotlb_start_ktime = ktime_to_ns(ktime_get()); 1362 1363 if ((type == QI_DIOTLB_TYPE || type == QI_DEIOTLB_TYPE) && 1364 dmar_latency_enabled(iommu, DMAR_LATENCY_INV_DEVTLB)) 1365 devtlb_start_ktime = ktime_to_ns(ktime_get()); 1366 1367 if (type == QI_IEC_TYPE && 1368 dmar_latency_enabled(iommu, DMAR_LATENCY_INV_IEC)) 1369 iec_start_ktime = ktime_to_ns(ktime_get()); 1370 1371 restart: 1372 rc = 0; 1373 1374 raw_spin_lock_irqsave(&qi->q_lock, flags); 1375 /* 1376 * Check if we have enough empty slots in the queue to submit, 1377 * the calculation is based on: 1378 * # of desc + 1 wait desc + 1 space between head and tail 1379 */ 1380 while (qi->free_cnt < count + 2) { 1381 raw_spin_unlock_irqrestore(&qi->q_lock, flags); 1382 cpu_relax(); 1383 raw_spin_lock_irqsave(&qi->q_lock, flags); 1384 } 1385 1386 index = qi->free_head; 1387 wait_index = (index + count) % QI_LENGTH; 1388 shift = qi_shift(iommu); 1389 1390 for (i = 0; i < count; i++) { 1391 offset = ((index + i) % QI_LENGTH) << shift; 1392 memcpy(qi->desc + offset, &desc[i], 1 << shift); 1393 qi->desc_status[(index + i) % QI_LENGTH] = QI_IN_USE; 1394 trace_qi_submit(iommu, desc[i].qw0, desc[i].qw1, 1395 desc[i].qw2, desc[i].qw3); 1396 } 1397 qi->desc_status[wait_index] = QI_IN_USE; 1398 1399 wait_desc.qw0 = QI_IWD_STATUS_DATA(QI_DONE) | 1400 QI_IWD_STATUS_WRITE | QI_IWD_TYPE; 1401 if (options & QI_OPT_WAIT_DRAIN) 1402 wait_desc.qw0 |= QI_IWD_PRQ_DRAIN; 1403 wait_desc.qw1 = virt_to_phys(&qi->desc_status[wait_index]); 1404 wait_desc.qw2 = 0; 1405 wait_desc.qw3 = 0; 1406 1407 offset = wait_index << shift; 1408 memcpy(qi->desc + offset, &wait_desc, 1 << shift); 1409 1410 qi->free_head = (qi->free_head + count + 1) % QI_LENGTH; 1411 qi->free_cnt -= count + 1; 1412 1413 /* 1414 * update the HW tail register indicating the presence of 1415 * new descriptors. 1416 */ 1417 writel(qi->free_head << shift, iommu->reg + DMAR_IQT_REG); 1418 1419 while (qi->desc_status[wait_index] != QI_DONE) { 1420 /* 1421 * We will leave the interrupts disabled, to prevent interrupt 1422 * context to queue another cmd while a cmd is already submitted 1423 * and waiting for completion on this cpu. This is to avoid 1424 * a deadlock where the interrupt context can wait indefinitely 1425 * for free slots in the queue. 1426 */ 1427 rc = qi_check_fault(iommu, index, wait_index); 1428 if (rc) 1429 break; 1430 1431 raw_spin_unlock(&qi->q_lock); 1432 cpu_relax(); 1433 raw_spin_lock(&qi->q_lock); 1434 } 1435 1436 for (i = 0; i < count; i++) 1437 qi->desc_status[(index + i) % QI_LENGTH] = QI_DONE; 1438 1439 reclaim_free_desc(qi); 1440 raw_spin_unlock_irqrestore(&qi->q_lock, flags); 1441 1442 if (rc == -EAGAIN) 1443 goto restart; 1444 1445 if (iotlb_start_ktime) 1446 dmar_latency_update(iommu, DMAR_LATENCY_INV_IOTLB, 1447 ktime_to_ns(ktime_get()) - iotlb_start_ktime); 1448 1449 if (devtlb_start_ktime) 1450 dmar_latency_update(iommu, DMAR_LATENCY_INV_DEVTLB, 1451 ktime_to_ns(ktime_get()) - devtlb_start_ktime); 1452 1453 if (iec_start_ktime) 1454 dmar_latency_update(iommu, DMAR_LATENCY_INV_IEC, 1455 ktime_to_ns(ktime_get()) - iec_start_ktime); 1456 1457 return rc; 1458 } 1459 1460 /* 1461 * Flush the global interrupt entry cache. 1462 */ 1463 void qi_global_iec(struct intel_iommu *iommu) 1464 { 1465 struct qi_desc desc; 1466 1467 desc.qw0 = QI_IEC_TYPE; 1468 desc.qw1 = 0; 1469 desc.qw2 = 0; 1470 desc.qw3 = 0; 1471 1472 /* should never fail */ 1473 qi_submit_sync(iommu, &desc, 1, 0); 1474 } 1475 1476 void qi_flush_context(struct intel_iommu *iommu, u16 did, u16 sid, u8 fm, 1477 u64 type) 1478 { 1479 struct qi_desc desc; 1480 1481 desc.qw0 = QI_CC_FM(fm) | QI_CC_SID(sid) | QI_CC_DID(did) 1482 | QI_CC_GRAN(type) | QI_CC_TYPE; 1483 desc.qw1 = 0; 1484 desc.qw2 = 0; 1485 desc.qw3 = 0; 1486 1487 qi_submit_sync(iommu, &desc, 1, 0); 1488 } 1489 1490 void qi_flush_iotlb(struct intel_iommu *iommu, u16 did, u64 addr, 1491 unsigned int size_order, u64 type) 1492 { 1493 u8 dw = 0, dr = 0; 1494 1495 struct qi_desc desc; 1496 int ih = 0; 1497 1498 if (cap_write_drain(iommu->cap)) 1499 dw = 1; 1500 1501 if (cap_read_drain(iommu->cap)) 1502 dr = 1; 1503 1504 desc.qw0 = QI_IOTLB_DID(did) | QI_IOTLB_DR(dr) | QI_IOTLB_DW(dw) 1505 | QI_IOTLB_GRAN(type) | QI_IOTLB_TYPE; 1506 desc.qw1 = QI_IOTLB_ADDR(addr) | QI_IOTLB_IH(ih) 1507 | QI_IOTLB_AM(size_order); 1508 desc.qw2 = 0; 1509 desc.qw3 = 0; 1510 1511 qi_submit_sync(iommu, &desc, 1, 0); 1512 } 1513 1514 void qi_flush_dev_iotlb(struct intel_iommu *iommu, u16 sid, u16 pfsid, 1515 u16 qdep, u64 addr, unsigned mask) 1516 { 1517 struct qi_desc desc; 1518 1519 if (mask) { 1520 addr |= (1ULL << (VTD_PAGE_SHIFT + mask - 1)) - 1; 1521 desc.qw1 = QI_DEV_IOTLB_ADDR(addr) | QI_DEV_IOTLB_SIZE; 1522 } else 1523 desc.qw1 = QI_DEV_IOTLB_ADDR(addr); 1524 1525 if (qdep >= QI_DEV_IOTLB_MAX_INVS) 1526 qdep = 0; 1527 1528 desc.qw0 = QI_DEV_IOTLB_SID(sid) | QI_DEV_IOTLB_QDEP(qdep) | 1529 QI_DIOTLB_TYPE | QI_DEV_IOTLB_PFSID(pfsid); 1530 desc.qw2 = 0; 1531 desc.qw3 = 0; 1532 1533 qi_submit_sync(iommu, &desc, 1, 0); 1534 } 1535 1536 /* PASID-based IOTLB invalidation */ 1537 void qi_flush_piotlb(struct intel_iommu *iommu, u16 did, u32 pasid, u64 addr, 1538 unsigned long npages, bool ih) 1539 { 1540 struct qi_desc desc = {.qw2 = 0, .qw3 = 0}; 1541 1542 /* 1543 * npages == -1 means a PASID-selective invalidation, otherwise, 1544 * a positive value for Page-selective-within-PASID invalidation. 1545 * 0 is not a valid input. 1546 */ 1547 if (WARN_ON(!npages)) { 1548 pr_err("Invalid input npages = %ld\n", npages); 1549 return; 1550 } 1551 1552 if (npages == -1) { 1553 desc.qw0 = QI_EIOTLB_PASID(pasid) | 1554 QI_EIOTLB_DID(did) | 1555 QI_EIOTLB_GRAN(QI_GRAN_NONG_PASID) | 1556 QI_EIOTLB_TYPE; 1557 desc.qw1 = 0; 1558 } else { 1559 int mask = ilog2(__roundup_pow_of_two(npages)); 1560 unsigned long align = (1ULL << (VTD_PAGE_SHIFT + mask)); 1561 1562 if (WARN_ON_ONCE(!IS_ALIGNED(addr, align))) 1563 addr = ALIGN_DOWN(addr, align); 1564 1565 desc.qw0 = QI_EIOTLB_PASID(pasid) | 1566 QI_EIOTLB_DID(did) | 1567 QI_EIOTLB_GRAN(QI_GRAN_PSI_PASID) | 1568 QI_EIOTLB_TYPE; 1569 desc.qw1 = QI_EIOTLB_ADDR(addr) | 1570 QI_EIOTLB_IH(ih) | 1571 QI_EIOTLB_AM(mask); 1572 } 1573 1574 qi_submit_sync(iommu, &desc, 1, 0); 1575 } 1576 1577 /* PASID-based device IOTLB Invalidate */ 1578 void qi_flush_dev_iotlb_pasid(struct intel_iommu *iommu, u16 sid, u16 pfsid, 1579 u32 pasid, u16 qdep, u64 addr, unsigned int size_order) 1580 { 1581 unsigned long mask = 1UL << (VTD_PAGE_SHIFT + size_order - 1); 1582 struct qi_desc desc = {.qw1 = 0, .qw2 = 0, .qw3 = 0}; 1583 1584 desc.qw0 = QI_DEV_EIOTLB_PASID(pasid) | QI_DEV_EIOTLB_SID(sid) | 1585 QI_DEV_EIOTLB_QDEP(qdep) | QI_DEIOTLB_TYPE | 1586 QI_DEV_IOTLB_PFSID(pfsid); 1587 1588 /* 1589 * If S bit is 0, we only flush a single page. If S bit is set, 1590 * The least significant zero bit indicates the invalidation address 1591 * range. VT-d spec 6.5.2.6. 1592 * e.g. address bit 12[0] indicates 8KB, 13[0] indicates 16KB. 1593 * size order = 0 is PAGE_SIZE 4KB 1594 * Max Invs Pending (MIP) is set to 0 for now until we have DIT in 1595 * ECAP. 1596 */ 1597 if (!IS_ALIGNED(addr, VTD_PAGE_SIZE << size_order)) 1598 pr_warn_ratelimited("Invalidate non-aligned address %llx, order %d\n", 1599 addr, size_order); 1600 1601 /* Take page address */ 1602 desc.qw1 = QI_DEV_EIOTLB_ADDR(addr); 1603 1604 if (size_order) { 1605 /* 1606 * Existing 0s in address below size_order may be the least 1607 * significant bit, we must set them to 1s to avoid having 1608 * smaller size than desired. 1609 */ 1610 desc.qw1 |= GENMASK_ULL(size_order + VTD_PAGE_SHIFT - 1, 1611 VTD_PAGE_SHIFT); 1612 /* Clear size_order bit to indicate size */ 1613 desc.qw1 &= ~mask; 1614 /* Set the S bit to indicate flushing more than 1 page */ 1615 desc.qw1 |= QI_DEV_EIOTLB_SIZE; 1616 } 1617 1618 qi_submit_sync(iommu, &desc, 1, 0); 1619 } 1620 1621 void qi_flush_pasid_cache(struct intel_iommu *iommu, u16 did, 1622 u64 granu, u32 pasid) 1623 { 1624 struct qi_desc desc = {.qw1 = 0, .qw2 = 0, .qw3 = 0}; 1625 1626 desc.qw0 = QI_PC_PASID(pasid) | QI_PC_DID(did) | 1627 QI_PC_GRAN(granu) | QI_PC_TYPE; 1628 qi_submit_sync(iommu, &desc, 1, 0); 1629 } 1630 1631 /* 1632 * Disable Queued Invalidation interface. 1633 */ 1634 void dmar_disable_qi(struct intel_iommu *iommu) 1635 { 1636 unsigned long flags; 1637 u32 sts; 1638 cycles_t start_time = get_cycles(); 1639 1640 if (!ecap_qis(iommu->ecap)) 1641 return; 1642 1643 raw_spin_lock_irqsave(&iommu->register_lock, flags); 1644 1645 sts = readl(iommu->reg + DMAR_GSTS_REG); 1646 if (!(sts & DMA_GSTS_QIES)) 1647 goto end; 1648 1649 /* 1650 * Give a chance to HW to complete the pending invalidation requests. 1651 */ 1652 while ((readl(iommu->reg + DMAR_IQT_REG) != 1653 readl(iommu->reg + DMAR_IQH_REG)) && 1654 (DMAR_OPERATION_TIMEOUT > (get_cycles() - start_time))) 1655 cpu_relax(); 1656 1657 iommu->gcmd &= ~DMA_GCMD_QIE; 1658 writel(iommu->gcmd, iommu->reg + DMAR_GCMD_REG); 1659 1660 IOMMU_WAIT_OP(iommu, DMAR_GSTS_REG, readl, 1661 !(sts & DMA_GSTS_QIES), sts); 1662 end: 1663 raw_spin_unlock_irqrestore(&iommu->register_lock, flags); 1664 } 1665 1666 /* 1667 * Enable queued invalidation. 1668 */ 1669 static void __dmar_enable_qi(struct intel_iommu *iommu) 1670 { 1671 u32 sts; 1672 unsigned long flags; 1673 struct q_inval *qi = iommu->qi; 1674 u64 val = virt_to_phys(qi->desc); 1675 1676 qi->free_head = qi->free_tail = 0; 1677 qi->free_cnt = QI_LENGTH; 1678 1679 /* 1680 * Set DW=1 and QS=1 in IQA_REG when Scalable Mode capability 1681 * is present. 1682 */ 1683 if (ecap_smts(iommu->ecap)) 1684 val |= (1 << 11) | 1; 1685 1686 raw_spin_lock_irqsave(&iommu->register_lock, flags); 1687 1688 /* write zero to the tail reg */ 1689 writel(0, iommu->reg + DMAR_IQT_REG); 1690 1691 dmar_writeq(iommu->reg + DMAR_IQA_REG, val); 1692 1693 iommu->gcmd |= DMA_GCMD_QIE; 1694 writel(iommu->gcmd, iommu->reg + DMAR_GCMD_REG); 1695 1696 /* Make sure hardware complete it */ 1697 IOMMU_WAIT_OP(iommu, DMAR_GSTS_REG, readl, (sts & DMA_GSTS_QIES), sts); 1698 1699 raw_spin_unlock_irqrestore(&iommu->register_lock, flags); 1700 } 1701 1702 /* 1703 * Enable Queued Invalidation interface. This is a must to support 1704 * interrupt-remapping. Also used by DMA-remapping, which replaces 1705 * register based IOTLB invalidation. 1706 */ 1707 int dmar_enable_qi(struct intel_iommu *iommu) 1708 { 1709 struct q_inval *qi; 1710 struct page *desc_page; 1711 1712 if (!ecap_qis(iommu->ecap)) 1713 return -ENOENT; 1714 1715 /* 1716 * queued invalidation is already setup and enabled. 1717 */ 1718 if (iommu->qi) 1719 return 0; 1720 1721 iommu->qi = kmalloc(sizeof(*qi), GFP_ATOMIC); 1722 if (!iommu->qi) 1723 return -ENOMEM; 1724 1725 qi = iommu->qi; 1726 1727 /* 1728 * Need two pages to accommodate 256 descriptors of 256 bits each 1729 * if the remapping hardware supports scalable mode translation. 1730 */ 1731 desc_page = alloc_pages_node(iommu->node, GFP_ATOMIC | __GFP_ZERO, 1732 !!ecap_smts(iommu->ecap)); 1733 if (!desc_page) { 1734 kfree(qi); 1735 iommu->qi = NULL; 1736 return -ENOMEM; 1737 } 1738 1739 qi->desc = page_address(desc_page); 1740 1741 qi->desc_status = kcalloc(QI_LENGTH, sizeof(int), GFP_ATOMIC); 1742 if (!qi->desc_status) { 1743 free_page((unsigned long) qi->desc); 1744 kfree(qi); 1745 iommu->qi = NULL; 1746 return -ENOMEM; 1747 } 1748 1749 raw_spin_lock_init(&qi->q_lock); 1750 1751 __dmar_enable_qi(iommu); 1752 1753 return 0; 1754 } 1755 1756 /* iommu interrupt handling. Most stuff are MSI-like. */ 1757 1758 enum faulttype { 1759 DMA_REMAP, 1760 INTR_REMAP, 1761 UNKNOWN, 1762 }; 1763 1764 static const char *dma_remap_fault_reasons[] = 1765 { 1766 "Software", 1767 "Present bit in root entry is clear", 1768 "Present bit in context entry is clear", 1769 "Invalid context entry", 1770 "Access beyond MGAW", 1771 "PTE Write access is not set", 1772 "PTE Read access is not set", 1773 "Next page table ptr is invalid", 1774 "Root table address invalid", 1775 "Context table ptr is invalid", 1776 "non-zero reserved fields in RTP", 1777 "non-zero reserved fields in CTP", 1778 "non-zero reserved fields in PTE", 1779 "PCE for translation request specifies blocking", 1780 }; 1781 1782 static const char * const dma_remap_sm_fault_reasons[] = { 1783 "SM: Invalid Root Table Address", 1784 "SM: TTM 0 for request with PASID", 1785 "SM: TTM 0 for page group request", 1786 "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", /* 0x33-0x37 */ 1787 "SM: Error attempting to access Root Entry", 1788 "SM: Present bit in Root Entry is clear", 1789 "SM: Non-zero reserved field set in Root Entry", 1790 "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", /* 0x3B-0x3F */ 1791 "SM: Error attempting to access Context Entry", 1792 "SM: Present bit in Context Entry is clear", 1793 "SM: Non-zero reserved field set in the Context Entry", 1794 "SM: Invalid Context Entry", 1795 "SM: DTE field in Context Entry is clear", 1796 "SM: PASID Enable field in Context Entry is clear", 1797 "SM: PASID is larger than the max in Context Entry", 1798 "SM: PRE field in Context-Entry is clear", 1799 "SM: RID_PASID field error in Context-Entry", 1800 "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", /* 0x49-0x4F */ 1801 "SM: Error attempting to access the PASID Directory Entry", 1802 "SM: Present bit in Directory Entry is clear", 1803 "SM: Non-zero reserved field set in PASID Directory Entry", 1804 "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", /* 0x53-0x57 */ 1805 "SM: Error attempting to access PASID Table Entry", 1806 "SM: Present bit in PASID Table Entry is clear", 1807 "SM: Non-zero reserved field set in PASID Table Entry", 1808 "SM: Invalid Scalable-Mode PASID Table Entry", 1809 "SM: ERE field is clear in PASID Table Entry", 1810 "SM: SRE field is clear in PASID Table Entry", 1811 "Unknown", "Unknown",/* 0x5E-0x5F */ 1812 "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", /* 0x60-0x67 */ 1813 "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", /* 0x68-0x6F */ 1814 "SM: Error attempting to access first-level paging entry", 1815 "SM: Present bit in first-level paging entry is clear", 1816 "SM: Non-zero reserved field set in first-level paging entry", 1817 "SM: Error attempting to access FL-PML4 entry", 1818 "SM: First-level entry address beyond MGAW in Nested translation", 1819 "SM: Read permission error in FL-PML4 entry in Nested translation", 1820 "SM: Read permission error in first-level paging entry in Nested translation", 1821 "SM: Write permission error in first-level paging entry in Nested translation", 1822 "SM: Error attempting to access second-level paging entry", 1823 "SM: Read/Write permission error in second-level paging entry", 1824 "SM: Non-zero reserved field set in second-level paging entry", 1825 "SM: Invalid second-level page table pointer", 1826 "SM: A/D bit update needed in second-level entry when set up in no snoop", 1827 "Unknown", "Unknown", "Unknown", /* 0x7D-0x7F */ 1828 "SM: Address in first-level translation is not canonical", 1829 "SM: U/S set 0 for first-level translation with user privilege", 1830 "SM: No execute permission for request with PASID and ER=1", 1831 "SM: Address beyond the DMA hardware max", 1832 "SM: Second-level entry address beyond the max", 1833 "SM: No write permission for Write/AtomicOp request", 1834 "SM: No read permission for Read/AtomicOp request", 1835 "SM: Invalid address-interrupt address", 1836 "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", /* 0x88-0x8F */ 1837 "SM: A/D bit update needed in first-level entry when set up in no snoop", 1838 }; 1839 1840 static const char *irq_remap_fault_reasons[] = 1841 { 1842 "Detected reserved fields in the decoded interrupt-remapped request", 1843 "Interrupt index exceeded the interrupt-remapping table size", 1844 "Present field in the IRTE entry is clear", 1845 "Error accessing interrupt-remapping table pointed by IRTA_REG", 1846 "Detected reserved fields in the IRTE entry", 1847 "Blocked a compatibility format interrupt request", 1848 "Blocked an interrupt request due to source-id verification failure", 1849 }; 1850 1851 static const char *dmar_get_fault_reason(u8 fault_reason, int *fault_type) 1852 { 1853 if (fault_reason >= 0x20 && (fault_reason - 0x20 < 1854 ARRAY_SIZE(irq_remap_fault_reasons))) { 1855 *fault_type = INTR_REMAP; 1856 return irq_remap_fault_reasons[fault_reason - 0x20]; 1857 } else if (fault_reason >= 0x30 && (fault_reason - 0x30 < 1858 ARRAY_SIZE(dma_remap_sm_fault_reasons))) { 1859 *fault_type = DMA_REMAP; 1860 return dma_remap_sm_fault_reasons[fault_reason - 0x30]; 1861 } else if (fault_reason < ARRAY_SIZE(dma_remap_fault_reasons)) { 1862 *fault_type = DMA_REMAP; 1863 return dma_remap_fault_reasons[fault_reason]; 1864 } else { 1865 *fault_type = UNKNOWN; 1866 return "Unknown"; 1867 } 1868 } 1869 1870 1871 static inline int dmar_msi_reg(struct intel_iommu *iommu, int irq) 1872 { 1873 if (iommu->irq == irq) 1874 return DMAR_FECTL_REG; 1875 else if (iommu->pr_irq == irq) 1876 return DMAR_PECTL_REG; 1877 else 1878 BUG(); 1879 } 1880 1881 void dmar_msi_unmask(struct irq_data *data) 1882 { 1883 struct intel_iommu *iommu = irq_data_get_irq_handler_data(data); 1884 int reg = dmar_msi_reg(iommu, data->irq); 1885 unsigned long flag; 1886 1887 /* unmask it */ 1888 raw_spin_lock_irqsave(&iommu->register_lock, flag); 1889 writel(0, iommu->reg + reg); 1890 /* Read a reg to force flush the post write */ 1891 readl(iommu->reg + reg); 1892 raw_spin_unlock_irqrestore(&iommu->register_lock, flag); 1893 } 1894 1895 void dmar_msi_mask(struct irq_data *data) 1896 { 1897 struct intel_iommu *iommu = irq_data_get_irq_handler_data(data); 1898 int reg = dmar_msi_reg(iommu, data->irq); 1899 unsigned long flag; 1900 1901 /* mask it */ 1902 raw_spin_lock_irqsave(&iommu->register_lock, flag); 1903 writel(DMA_FECTL_IM, iommu->reg + reg); 1904 /* Read a reg to force flush the post write */ 1905 readl(iommu->reg + reg); 1906 raw_spin_unlock_irqrestore(&iommu->register_lock, flag); 1907 } 1908 1909 void dmar_msi_write(int irq, struct msi_msg *msg) 1910 { 1911 struct intel_iommu *iommu = irq_get_handler_data(irq); 1912 int reg = dmar_msi_reg(iommu, irq); 1913 unsigned long flag; 1914 1915 raw_spin_lock_irqsave(&iommu->register_lock, flag); 1916 writel(msg->data, iommu->reg + reg + 4); 1917 writel(msg->address_lo, iommu->reg + reg + 8); 1918 writel(msg->address_hi, iommu->reg + reg + 12); 1919 raw_spin_unlock_irqrestore(&iommu->register_lock, flag); 1920 } 1921 1922 void dmar_msi_read(int irq, struct msi_msg *msg) 1923 { 1924 struct intel_iommu *iommu = irq_get_handler_data(irq); 1925 int reg = dmar_msi_reg(iommu, irq); 1926 unsigned long flag; 1927 1928 raw_spin_lock_irqsave(&iommu->register_lock, flag); 1929 msg->data = readl(iommu->reg + reg + 4); 1930 msg->address_lo = readl(iommu->reg + reg + 8); 1931 msg->address_hi = readl(iommu->reg + reg + 12); 1932 raw_spin_unlock_irqrestore(&iommu->register_lock, flag); 1933 } 1934 1935 static int dmar_fault_do_one(struct intel_iommu *iommu, int type, 1936 u8 fault_reason, u32 pasid, u16 source_id, 1937 unsigned long long addr) 1938 { 1939 const char *reason; 1940 int fault_type; 1941 1942 reason = dmar_get_fault_reason(fault_reason, &fault_type); 1943 1944 if (fault_type == INTR_REMAP) { 1945 pr_err("[INTR-REMAP] Request device [%02x:%02x.%d] fault index 0x%llx [fault reason 0x%02x] %s\n", 1946 source_id >> 8, PCI_SLOT(source_id & 0xFF), 1947 PCI_FUNC(source_id & 0xFF), addr >> 48, 1948 fault_reason, reason); 1949 1950 return 0; 1951 } 1952 1953 if (pasid == INVALID_IOASID) 1954 pr_err("[%s NO_PASID] Request device [%02x:%02x.%d] fault addr 0x%llx [fault reason 0x%02x] %s\n", 1955 type ? "DMA Read" : "DMA Write", 1956 source_id >> 8, PCI_SLOT(source_id & 0xFF), 1957 PCI_FUNC(source_id & 0xFF), addr, 1958 fault_reason, reason); 1959 else 1960 pr_err("[%s PASID 0x%x] Request device [%02x:%02x.%d] fault addr 0x%llx [fault reason 0x%02x] %s\n", 1961 type ? "DMA Read" : "DMA Write", pasid, 1962 source_id >> 8, PCI_SLOT(source_id & 0xFF), 1963 PCI_FUNC(source_id & 0xFF), addr, 1964 fault_reason, reason); 1965 1966 dmar_fault_dump_ptes(iommu, source_id, addr, pasid); 1967 1968 return 0; 1969 } 1970 1971 #define PRIMARY_FAULT_REG_LEN (16) 1972 irqreturn_t dmar_fault(int irq, void *dev_id) 1973 { 1974 struct intel_iommu *iommu = dev_id; 1975 int reg, fault_index; 1976 u32 fault_status; 1977 unsigned long flag; 1978 static DEFINE_RATELIMIT_STATE(rs, 1979 DEFAULT_RATELIMIT_INTERVAL, 1980 DEFAULT_RATELIMIT_BURST); 1981 1982 raw_spin_lock_irqsave(&iommu->register_lock, flag); 1983 fault_status = readl(iommu->reg + DMAR_FSTS_REG); 1984 if (fault_status && __ratelimit(&rs)) 1985 pr_err("DRHD: handling fault status reg %x\n", fault_status); 1986 1987 /* TBD: ignore advanced fault log currently */ 1988 if (!(fault_status & DMA_FSTS_PPF)) 1989 goto unlock_exit; 1990 1991 fault_index = dma_fsts_fault_record_index(fault_status); 1992 reg = cap_fault_reg_offset(iommu->cap); 1993 while (1) { 1994 /* Disable printing, simply clear the fault when ratelimited */ 1995 bool ratelimited = !__ratelimit(&rs); 1996 u8 fault_reason; 1997 u16 source_id; 1998 u64 guest_addr; 1999 u32 pasid; 2000 int type; 2001 u32 data; 2002 bool pasid_present; 2003 2004 /* highest 32 bits */ 2005 data = readl(iommu->reg + reg + 2006 fault_index * PRIMARY_FAULT_REG_LEN + 12); 2007 if (!(data & DMA_FRCD_F)) 2008 break; 2009 2010 if (!ratelimited) { 2011 fault_reason = dma_frcd_fault_reason(data); 2012 type = dma_frcd_type(data); 2013 2014 pasid = dma_frcd_pasid_value(data); 2015 data = readl(iommu->reg + reg + 2016 fault_index * PRIMARY_FAULT_REG_LEN + 8); 2017 source_id = dma_frcd_source_id(data); 2018 2019 pasid_present = dma_frcd_pasid_present(data); 2020 guest_addr = dmar_readq(iommu->reg + reg + 2021 fault_index * PRIMARY_FAULT_REG_LEN); 2022 guest_addr = dma_frcd_page_addr(guest_addr); 2023 } 2024 2025 /* clear the fault */ 2026 writel(DMA_FRCD_F, iommu->reg + reg + 2027 fault_index * PRIMARY_FAULT_REG_LEN + 12); 2028 2029 raw_spin_unlock_irqrestore(&iommu->register_lock, flag); 2030 2031 if (!ratelimited) 2032 /* Using pasid -1 if pasid is not present */ 2033 dmar_fault_do_one(iommu, type, fault_reason, 2034 pasid_present ? pasid : INVALID_IOASID, 2035 source_id, guest_addr); 2036 2037 fault_index++; 2038 if (fault_index >= cap_num_fault_regs(iommu->cap)) 2039 fault_index = 0; 2040 raw_spin_lock_irqsave(&iommu->register_lock, flag); 2041 } 2042 2043 writel(DMA_FSTS_PFO | DMA_FSTS_PPF | DMA_FSTS_PRO, 2044 iommu->reg + DMAR_FSTS_REG); 2045 2046 unlock_exit: 2047 raw_spin_unlock_irqrestore(&iommu->register_lock, flag); 2048 return IRQ_HANDLED; 2049 } 2050 2051 int dmar_set_interrupt(struct intel_iommu *iommu) 2052 { 2053 int irq, ret; 2054 2055 /* 2056 * Check if the fault interrupt is already initialized. 2057 */ 2058 if (iommu->irq) 2059 return 0; 2060 2061 irq = dmar_alloc_hwirq(iommu->seq_id, iommu->node, iommu); 2062 if (irq > 0) { 2063 iommu->irq = irq; 2064 } else { 2065 pr_err("No free IRQ vectors\n"); 2066 return -EINVAL; 2067 } 2068 2069 ret = request_irq(irq, dmar_fault, IRQF_NO_THREAD, iommu->name, iommu); 2070 if (ret) 2071 pr_err("Can't request irq\n"); 2072 return ret; 2073 } 2074 2075 int __init enable_drhd_fault_handling(void) 2076 { 2077 struct dmar_drhd_unit *drhd; 2078 struct intel_iommu *iommu; 2079 2080 /* 2081 * Enable fault control interrupt. 2082 */ 2083 for_each_iommu(iommu, drhd) { 2084 u32 fault_status; 2085 int ret = dmar_set_interrupt(iommu); 2086 2087 if (ret) { 2088 pr_err("DRHD %Lx: failed to enable fault, interrupt, ret %d\n", 2089 (unsigned long long)drhd->reg_base_addr, ret); 2090 return -1; 2091 } 2092 2093 /* 2094 * Clear any previous faults. 2095 */ 2096 dmar_fault(iommu->irq, iommu); 2097 fault_status = readl(iommu->reg + DMAR_FSTS_REG); 2098 writel(fault_status, iommu->reg + DMAR_FSTS_REG); 2099 } 2100 2101 return 0; 2102 } 2103 2104 /* 2105 * Re-enable Queued Invalidation interface. 2106 */ 2107 int dmar_reenable_qi(struct intel_iommu *iommu) 2108 { 2109 if (!ecap_qis(iommu->ecap)) 2110 return -ENOENT; 2111 2112 if (!iommu->qi) 2113 return -ENOENT; 2114 2115 /* 2116 * First disable queued invalidation. 2117 */ 2118 dmar_disable_qi(iommu); 2119 /* 2120 * Then enable queued invalidation again. Since there is no pending 2121 * invalidation requests now, it's safe to re-enable queued 2122 * invalidation. 2123 */ 2124 __dmar_enable_qi(iommu); 2125 2126 return 0; 2127 } 2128 2129 /* 2130 * Check interrupt remapping support in DMAR table description. 2131 */ 2132 int __init dmar_ir_support(void) 2133 { 2134 struct acpi_table_dmar *dmar; 2135 dmar = (struct acpi_table_dmar *)dmar_tbl; 2136 if (!dmar) 2137 return 0; 2138 return dmar->flags & 0x1; 2139 } 2140 2141 /* Check whether DMAR units are in use */ 2142 static inline bool dmar_in_use(void) 2143 { 2144 return irq_remapping_enabled || intel_iommu_enabled; 2145 } 2146 2147 static int __init dmar_free_unused_resources(void) 2148 { 2149 struct dmar_drhd_unit *dmaru, *dmaru_n; 2150 2151 if (dmar_in_use()) 2152 return 0; 2153 2154 if (dmar_dev_scope_status != 1 && !list_empty(&dmar_drhd_units)) 2155 bus_unregister_notifier(&pci_bus_type, &dmar_pci_bus_nb); 2156 2157 down_write(&dmar_global_lock); 2158 list_for_each_entry_safe(dmaru, dmaru_n, &dmar_drhd_units, list) { 2159 list_del(&dmaru->list); 2160 dmar_free_drhd(dmaru); 2161 } 2162 up_write(&dmar_global_lock); 2163 2164 return 0; 2165 } 2166 2167 late_initcall(dmar_free_unused_resources); 2168 IOMMU_INIT_POST(detect_intel_iommu); 2169 2170 /* 2171 * DMAR Hotplug Support 2172 * For more details, please refer to Intel(R) Virtualization Technology 2173 * for Directed-IO Architecture Specifiction, Rev 2.2, Section 8.8 2174 * "Remapping Hardware Unit Hot Plug". 2175 */ 2176 static guid_t dmar_hp_guid = 2177 GUID_INIT(0xD8C1A3A6, 0xBE9B, 0x4C9B, 2178 0x91, 0xBF, 0xC3, 0xCB, 0x81, 0xFC, 0x5D, 0xAF); 2179 2180 /* 2181 * Currently there's only one revision and BIOS will not check the revision id, 2182 * so use 0 for safety. 2183 */ 2184 #define DMAR_DSM_REV_ID 0 2185 #define DMAR_DSM_FUNC_DRHD 1 2186 #define DMAR_DSM_FUNC_ATSR 2 2187 #define DMAR_DSM_FUNC_RHSA 3 2188 #define DMAR_DSM_FUNC_SATC 4 2189 2190 static inline bool dmar_detect_dsm(acpi_handle handle, int func) 2191 { 2192 return acpi_check_dsm(handle, &dmar_hp_guid, DMAR_DSM_REV_ID, 1 << func); 2193 } 2194 2195 static int dmar_walk_dsm_resource(acpi_handle handle, int func, 2196 dmar_res_handler_t handler, void *arg) 2197 { 2198 int ret = -ENODEV; 2199 union acpi_object *obj; 2200 struct acpi_dmar_header *start; 2201 struct dmar_res_callback callback; 2202 static int res_type[] = { 2203 [DMAR_DSM_FUNC_DRHD] = ACPI_DMAR_TYPE_HARDWARE_UNIT, 2204 [DMAR_DSM_FUNC_ATSR] = ACPI_DMAR_TYPE_ROOT_ATS, 2205 [DMAR_DSM_FUNC_RHSA] = ACPI_DMAR_TYPE_HARDWARE_AFFINITY, 2206 [DMAR_DSM_FUNC_SATC] = ACPI_DMAR_TYPE_SATC, 2207 }; 2208 2209 if (!dmar_detect_dsm(handle, func)) 2210 return 0; 2211 2212 obj = acpi_evaluate_dsm_typed(handle, &dmar_hp_guid, DMAR_DSM_REV_ID, 2213 func, NULL, ACPI_TYPE_BUFFER); 2214 if (!obj) 2215 return -ENODEV; 2216 2217 memset(&callback, 0, sizeof(callback)); 2218 callback.cb[res_type[func]] = handler; 2219 callback.arg[res_type[func]] = arg; 2220 start = (struct acpi_dmar_header *)obj->buffer.pointer; 2221 ret = dmar_walk_remapping_entries(start, obj->buffer.length, &callback); 2222 2223 ACPI_FREE(obj); 2224 2225 return ret; 2226 } 2227 2228 static int dmar_hp_add_drhd(struct acpi_dmar_header *header, void *arg) 2229 { 2230 int ret; 2231 struct dmar_drhd_unit *dmaru; 2232 2233 dmaru = dmar_find_dmaru((struct acpi_dmar_hardware_unit *)header); 2234 if (!dmaru) 2235 return -ENODEV; 2236 2237 ret = dmar_ir_hotplug(dmaru, true); 2238 if (ret == 0) 2239 ret = dmar_iommu_hotplug(dmaru, true); 2240 2241 return ret; 2242 } 2243 2244 static int dmar_hp_remove_drhd(struct acpi_dmar_header *header, void *arg) 2245 { 2246 int i, ret; 2247 struct device *dev; 2248 struct dmar_drhd_unit *dmaru; 2249 2250 dmaru = dmar_find_dmaru((struct acpi_dmar_hardware_unit *)header); 2251 if (!dmaru) 2252 return 0; 2253 2254 /* 2255 * All PCI devices managed by this unit should have been destroyed. 2256 */ 2257 if (!dmaru->include_all && dmaru->devices && dmaru->devices_cnt) { 2258 for_each_active_dev_scope(dmaru->devices, 2259 dmaru->devices_cnt, i, dev) 2260 return -EBUSY; 2261 } 2262 2263 ret = dmar_ir_hotplug(dmaru, false); 2264 if (ret == 0) 2265 ret = dmar_iommu_hotplug(dmaru, false); 2266 2267 return ret; 2268 } 2269 2270 static int dmar_hp_release_drhd(struct acpi_dmar_header *header, void *arg) 2271 { 2272 struct dmar_drhd_unit *dmaru; 2273 2274 dmaru = dmar_find_dmaru((struct acpi_dmar_hardware_unit *)header); 2275 if (dmaru) { 2276 list_del_rcu(&dmaru->list); 2277 synchronize_rcu(); 2278 dmar_free_drhd(dmaru); 2279 } 2280 2281 return 0; 2282 } 2283 2284 static int dmar_hotplug_insert(acpi_handle handle) 2285 { 2286 int ret; 2287 int drhd_count = 0; 2288 2289 ret = dmar_walk_dsm_resource(handle, DMAR_DSM_FUNC_DRHD, 2290 &dmar_validate_one_drhd, (void *)1); 2291 if (ret) 2292 goto out; 2293 2294 ret = dmar_walk_dsm_resource(handle, DMAR_DSM_FUNC_DRHD, 2295 &dmar_parse_one_drhd, (void *)&drhd_count); 2296 if (ret == 0 && drhd_count == 0) { 2297 pr_warn(FW_BUG "No DRHD structures in buffer returned by _DSM method\n"); 2298 goto out; 2299 } else if (ret) { 2300 goto release_drhd; 2301 } 2302 2303 ret = dmar_walk_dsm_resource(handle, DMAR_DSM_FUNC_RHSA, 2304 &dmar_parse_one_rhsa, NULL); 2305 if (ret) 2306 goto release_drhd; 2307 2308 ret = dmar_walk_dsm_resource(handle, DMAR_DSM_FUNC_ATSR, 2309 &dmar_parse_one_atsr, NULL); 2310 if (ret) 2311 goto release_atsr; 2312 2313 ret = dmar_walk_dsm_resource(handle, DMAR_DSM_FUNC_DRHD, 2314 &dmar_hp_add_drhd, NULL); 2315 if (!ret) 2316 return 0; 2317 2318 dmar_walk_dsm_resource(handle, DMAR_DSM_FUNC_DRHD, 2319 &dmar_hp_remove_drhd, NULL); 2320 release_atsr: 2321 dmar_walk_dsm_resource(handle, DMAR_DSM_FUNC_ATSR, 2322 &dmar_release_one_atsr, NULL); 2323 release_drhd: 2324 dmar_walk_dsm_resource(handle, DMAR_DSM_FUNC_DRHD, 2325 &dmar_hp_release_drhd, NULL); 2326 out: 2327 return ret; 2328 } 2329 2330 static int dmar_hotplug_remove(acpi_handle handle) 2331 { 2332 int ret; 2333 2334 ret = dmar_walk_dsm_resource(handle, DMAR_DSM_FUNC_ATSR, 2335 &dmar_check_one_atsr, NULL); 2336 if (ret) 2337 return ret; 2338 2339 ret = dmar_walk_dsm_resource(handle, DMAR_DSM_FUNC_DRHD, 2340 &dmar_hp_remove_drhd, NULL); 2341 if (ret == 0) { 2342 WARN_ON(dmar_walk_dsm_resource(handle, DMAR_DSM_FUNC_ATSR, 2343 &dmar_release_one_atsr, NULL)); 2344 WARN_ON(dmar_walk_dsm_resource(handle, DMAR_DSM_FUNC_DRHD, 2345 &dmar_hp_release_drhd, NULL)); 2346 } else { 2347 dmar_walk_dsm_resource(handle, DMAR_DSM_FUNC_DRHD, 2348 &dmar_hp_add_drhd, NULL); 2349 } 2350 2351 return ret; 2352 } 2353 2354 static acpi_status dmar_get_dsm_handle(acpi_handle handle, u32 lvl, 2355 void *context, void **retval) 2356 { 2357 acpi_handle *phdl = retval; 2358 2359 if (dmar_detect_dsm(handle, DMAR_DSM_FUNC_DRHD)) { 2360 *phdl = handle; 2361 return AE_CTRL_TERMINATE; 2362 } 2363 2364 return AE_OK; 2365 } 2366 2367 static int dmar_device_hotplug(acpi_handle handle, bool insert) 2368 { 2369 int ret; 2370 acpi_handle tmp = NULL; 2371 acpi_status status; 2372 2373 if (!dmar_in_use()) 2374 return 0; 2375 2376 if (dmar_detect_dsm(handle, DMAR_DSM_FUNC_DRHD)) { 2377 tmp = handle; 2378 } else { 2379 status = acpi_walk_namespace(ACPI_TYPE_DEVICE, handle, 2380 ACPI_UINT32_MAX, 2381 dmar_get_dsm_handle, 2382 NULL, NULL, &tmp); 2383 if (ACPI_FAILURE(status)) { 2384 pr_warn("Failed to locate _DSM method.\n"); 2385 return -ENXIO; 2386 } 2387 } 2388 if (tmp == NULL) 2389 return 0; 2390 2391 down_write(&dmar_global_lock); 2392 if (insert) 2393 ret = dmar_hotplug_insert(tmp); 2394 else 2395 ret = dmar_hotplug_remove(tmp); 2396 up_write(&dmar_global_lock); 2397 2398 return ret; 2399 } 2400 2401 int dmar_device_add(acpi_handle handle) 2402 { 2403 return dmar_device_hotplug(handle, true); 2404 } 2405 2406 int dmar_device_remove(acpi_handle handle) 2407 { 2408 return dmar_device_hotplug(handle, false); 2409 } 2410 2411 /* 2412 * dmar_platform_optin - Is %DMA_CTRL_PLATFORM_OPT_IN_FLAG set in DMAR table 2413 * 2414 * Returns true if the platform has %DMA_CTRL_PLATFORM_OPT_IN_FLAG set in 2415 * the ACPI DMAR table. This means that the platform boot firmware has made 2416 * sure no device can issue DMA outside of RMRR regions. 2417 */ 2418 bool dmar_platform_optin(void) 2419 { 2420 struct acpi_table_dmar *dmar; 2421 acpi_status status; 2422 bool ret; 2423 2424 status = acpi_get_table(ACPI_SIG_DMAR, 0, 2425 (struct acpi_table_header **)&dmar); 2426 if (ACPI_FAILURE(status)) 2427 return false; 2428 2429 ret = !!(dmar->flags & DMAR_PLATFORM_OPT_IN); 2430 acpi_put_table((struct acpi_table_header *)dmar); 2431 2432 return ret; 2433 } 2434 EXPORT_SYMBOL_GPL(dmar_platform_optin); 2435