drivers/pci/dmar.c

   1 /*
   2  * Copyright (c) 2006, Intel Corporation.
   3  *
   4  * This program is free software; you can redistribute it and/or modify it
   5  * under the terms and conditions of the GNU General Public License,
   6  * version 2, as published by the Free Software Foundation.
   7  *
   8  * This program is distributed in the hope it will be useful, but WITHOUT
   9  * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
  10  * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
  11  * more details.
  12  *
  13  * You should have received a copy of the GNU General Public License along with
  14  * this program; if not, write to the Free Software Foundation, Inc., 59 Temple
  15  * Place - Suite 330, Boston, MA 02111-1307 USA.
  16  *
  17  * Copyright (C) 2006-2008 Intel Corporation
  18  * Author: Ashok Raj <ashok.raj@intel.com>
  19  * Author: Shaohua Li <shaohua.li@intel.com>
  20  * Author: Anil S Keshavamurthy <anil.s.keshavamurthy@intel.com>
  21  *
  22  * This file implements early detection/parsing of Remapping Devices
  23  * reported to OS through BIOS via DMA remapping reporting (DMAR) ACPI
  24  * tables.
  25  *
  26  * These routines are used by both DMA-remapping and Interrupt-remapping
  27  */
  28
  29 #include <linux/pci.h>
  30 #include <linux/dmar.h>
  31 #include <linux/iova.h>
  32 #include <linux/intel-iommu.h>
  33 #include <linux/timer.h>
  34 #include <linux/irq.h>
  35 #include <linux/interrupt.h>
  36 #include <linux/tboot.h>
  37 #include <linux/dmi.h>
  38 #include <linux/slab.h>
  39
  40 #define PREFIX "DMAR: "
  41
  42 /* No locks are needed as DMA remapping hardware unit
  43  * list is constructed at boot time and hotplug of
  44  * these units are not supported by the architecture.
  45  */
  46 LIST_HEAD(dmar_drhd_units);
  47
  48 static struct acpi_table_header * __initdata dmar_tbl;
  49 static acpi_size dmar_tbl_size;
  50
  51 static void __init dmar_register_drhd_unit(struct dmar_drhd_unit *drhd)
  52 {
  53         /*
  54          * add INCLUDE_ALL at the tail, so scan the list will find it at
  55          * the very end.
  56          */
  57         if (drhd->include_all)
  58                 list_add_tail(&drhd->list, &dmar_drhd_units);
  59         else
  60                 list_add(&drhd->list, &dmar_drhd_units);
  61 }
  62
  63 static int __init dmar_parse_one_dev_scope(struct acpi_dmar_device_scope *scope,
  64                                            struct pci_dev **dev, u16 segment)
  65 {
  66         struct pci_bus *bus;
  67         struct pci_dev *pdev = NULL;
  68         struct acpi_dmar_pci_path *path;
  69         int count;
  70
  71         bus = pci_find_bus(segment, scope->bus);
  72         path = (struct acpi_dmar_pci_path *)(scope + 1);
  73         count = (scope->length - sizeof(struct acpi_dmar_device_scope))
  74                 / sizeof(struct acpi_dmar_pci_path);
  75
  76         while (count) {
  77                 if (pdev)
  78                         pci_dev_put(pdev);
  79                 /*
  80                  * Some BIOSes list non-exist devices in DMAR table, just
  81                  * ignore it
  82                  */
  83                 if (!bus) {
  84                         printk(KERN_WARNING
  85                         PREFIX "Device scope bus [%d] not found\n",
  86                         scope->bus);
  87                         break;
  88                 }
  89                 pdev = pci_get_slot(bus, PCI_DEVFN(path->dev, path->fn));
  90                 if (!pdev) {
  91                         printk(KERN_WARNING PREFIX
  92                         "Device scope device [%04x:%02x:%02x.%02x] not found\n",
  93                                 segment, bus->number, path->dev, path->fn);
  94                         break;
  95                 }
  96                 path ++;
  97                 count --;
  98                 bus = pdev->subordinate;
  99         }
 100         if (!pdev) {
 101                 printk(KERN_WARNING PREFIX
 102                 "Device scope device [%04x:%02x:%02x.%02x] not found\n",
 103                 segment, scope->bus, path->dev, path->fn);
 104                 *dev = NULL;
 105                 return 0;
 106         }
 107         if ((scope->entry_type == ACPI_DMAR_SCOPE_TYPE_ENDPOINT && \
 108                         pdev->subordinate) || (scope->entry_type == \
 109                         ACPI_DMAR_SCOPE_TYPE_BRIDGE && !pdev->subordinate)) {
 110                 pci_dev_put(pdev);
 111                 printk(KERN_WARNING PREFIX
 112                         "Device scope type does not match for %s\n",
 113                          pci_name(pdev));
 114                 return -EINVAL;
 115         }
 116         *dev = pdev;
 117         return 0;
 118 }
 119
 120 static int __init dmar_parse_dev_scope(void *start, void *end, int *cnt,
 121                                        struct pci_dev ***devices, u16 segment)
 122 {
 123         struct acpi_dmar_device_scope *scope;
 124         void * tmp = start;
 125         int index;
 126         int ret;
 127
 128         *cnt = 0;
 129         while (start < end) {
 130                 scope = start;
 131                 if (scope->entry_type == ACPI_DMAR_SCOPE_TYPE_ENDPOINT ||
 132                     scope->entry_type == ACPI_DMAR_SCOPE_TYPE_BRIDGE)
 133                         (*cnt)++;
 134                 else if (scope->entry_type != ACPI_DMAR_SCOPE_TYPE_IOAPIC) {
 135                         printk(KERN_WARNING PREFIX
 136                                "Unsupported device scope\n");
 137                 }
 138                 start += scope->length;
 139         }
 140         if (*cnt == 0)
 141                 return 0;
 142
 143         *devices = kcalloc(*cnt, sizeof(struct pci_dev *), GFP_KERNEL);
 144         if (!*devices)
 145                 return -ENOMEM;
 146
 147         start = tmp;
 148         index = 0;
 149         while (start < end) {
 150                 scope = start;
 151                 if (scope->entry_type == ACPI_DMAR_SCOPE_TYPE_ENDPOINT ||
 152                     scope->entry_type == ACPI_DMAR_SCOPE_TYPE_BRIDGE) {
 153                         ret = dmar_parse_one_dev_scope(scope,
 154                                 &(*devices)[index], segment);
 155                         if (ret) {
 156                                 kfree(*devices);
 157                                 return ret;
 158                         }
 159                         index ++;
 160                 }
 161                 start += scope->length;
 162         }
 163
 164         return 0;
 165 }
 166
 167 /**
 168  * dmar_parse_one_drhd - parses exactly one DMA remapping hardware definition
 169  * structure which uniquely represent one DMA remapping hardware unit
 170  * present in the platform
 171  */
 172 static int __init
 173 dmar_parse_one_drhd(struct acpi_dmar_header *header)
 174 {
 175         struct acpi_dmar_hardware_unit *drhd;
 176         struct dmar_drhd_unit *dmaru;
 177         int ret = 0;
 178
 179         drhd = (struct acpi_dmar_hardware_unit *)header;
 180         dmaru = kzalloc(sizeof(*dmaru), GFP_KERNEL);
 181         if (!dmaru)
 182                 return -ENOMEM;
 183
 184         dmaru->hdr = header;
 185         dmaru->reg_base_addr = drhd->address;
 186         dmaru->segment = drhd->segment;
 187         dmaru->include_all = drhd->flags & 0x1; /* BIT0: INCLUDE_ALL */
 188
 189         ret = alloc_iommu(dmaru);
 190         if (ret) {
 191                 kfree(dmaru);
 192                 return ret;
 193         }
 194         dmar_register_drhd_unit(dmaru);
 195         return 0;
 196 }
 197
 198 static int __init dmar_parse_dev(struct dmar_drhd_unit *dmaru)
 199 {
 200         struct acpi_dmar_hardware_unit *drhd;
 201         int ret = 0;
 202
 203         drhd = (struct acpi_dmar_hardware_unit *) dmaru->hdr;
 204
 205         if (dmaru->include_all)
 206                 return 0;
 207
 208         ret = dmar_parse_dev_scope((void *)(drhd + 1),
 209                                 ((void *)drhd) + drhd->header.length,
 210                                 &dmaru->devices_cnt, &dmaru->devices,
 211                                 drhd->segment);
 212         if (ret) {
 213                 list_del(&dmaru->list);
 214                 kfree(dmaru);
 215         }
 216         return ret;
 217 }
 218
 219 #ifdef CONFIG_DMAR
 220 LIST_HEAD(dmar_rmrr_units);
 221
 222 static void __init dmar_register_rmrr_unit(struct dmar_rmrr_unit *rmrr)
 223 {
 224         list_add(&rmrr->list, &dmar_rmrr_units);
 225 }
 226
 227
 228 static int __init
 229 dmar_parse_one_rmrr(struct acpi_dmar_header *header)
 230 {
 231         struct acpi_dmar_reserved_memory *rmrr;
 232         struct dmar_rmrr_unit *rmrru;
 233
 234         rmrru = kzalloc(sizeof(*rmrru), GFP_KERNEL);
 235         if (!rmrru)
 236                 return -ENOMEM;
 237
 238         rmrru->hdr = header;
 239         rmrr = (struct acpi_dmar_reserved_memory *)header;
 240         rmrru->base_address = rmrr->base_address;
 241         rmrru->end_address = rmrr->end_address;
 242
 243         dmar_register_rmrr_unit(rmrru);
 244         return 0;
 245 }
 246
 247 static int __init
 248 rmrr_parse_dev(struct dmar_rmrr_unit *rmrru)
 249 {
 250         struct acpi_dmar_reserved_memory *rmrr;
 251         int ret;
 252
 253         rmrr = (struct acpi_dmar_reserved_memory *) rmrru->hdr;
 254         ret = dmar_parse_dev_scope((void *)(rmrr + 1),
 255                 ((void *)rmrr) + rmrr->header.length,
 256                 &rmrru->devices_cnt, &rmrru->devices, rmrr->segment);
 257
 258         if (ret || (rmrru->devices_cnt == 0)) {
 259                 list_del(&rmrru->list);
 260                 kfree(rmrru);
 261         }
 262         return ret;
 263 }
 264
 265 static LIST_HEAD(dmar_atsr_units);
 266
 267 static int __init dmar_parse_one_atsr(struct acpi_dmar_header *hdr)
 268 {
 269         struct acpi_dmar_atsr *atsr;
 270         struct dmar_atsr_unit *atsru;
 271
 272         atsr = container_of(hdr, struct acpi_dmar_atsr, header);
 273         atsru = kzalloc(sizeof(*atsru), GFP_KERNEL);
 274         if (!atsru)
 275                 return -ENOMEM;
 276
 277         atsru->hdr = hdr;
 278         atsru->include_all = atsr->flags & 0x1;
 279
 280         list_add(&atsru->list, &dmar_atsr_units);
 281
 282         return 0;
 283 }
 284
 285 static int __init atsr_parse_dev(struct dmar_atsr_unit *atsru)
 286 {
 287         int rc;
 288         struct acpi_dmar_atsr *atsr;
 289
 290         if (atsru->include_all)
 291                 return 0;
 292
 293         atsr = container_of(atsru->hdr, struct acpi_dmar_atsr, header);
 294         rc = dmar_parse_dev_scope((void *)(atsr + 1),
 295                                 (void *)atsr + atsr->header.length,
 296                                 &atsru->devices_cnt, &atsru->devices,
 297                                 atsr->segment);
 298         if (rc || !atsru->devices_cnt) {
 299                 list_del(&atsru->list);
 300                 kfree(atsru);
 301         }
 302
 303         return rc;
 304 }
 305
 306 int dmar_find_matched_atsr_unit(struct pci_dev *dev)
 307 {
 308         int i;
 309         struct pci_bus *bus;
 310         struct acpi_dmar_atsr *atsr;
 311         struct dmar_atsr_unit *atsru;
 312
 313         dev = pci_physfn(dev);
 314
 315         list_for_each_entry(atsru, &dmar_atsr_units, list) {
 316                 atsr = container_of(atsru->hdr, struct acpi_dmar_atsr, header);
 317                 if (atsr->segment == pci_domain_nr(dev->bus))
 318                         goto found;
 319         }
 320
 321         return 0;
 322
 323 found:
 324         for (bus = dev->bus; bus; bus = bus->parent) {
 325                 struct pci_dev *bridge = bus->self;
 326
 327                 if (!bridge || !pci_is_pcie(bridge) ||
 328                     bridge->pcie_type == PCI_EXP_TYPE_PCI_BRIDGE)
 329                         return 0;
 330
 331                 if (bridge->pcie_type == PCI_EXP_TYPE_ROOT_PORT) {
 332                         for (i = 0; i < atsru->devices_cnt; i++)
 333                                 if (atsru->devices[i] == bridge)
 334                                         return 1;
 335                         break;
 336                 }
 337         }
 338
 339         if (atsru->include_all)
 340                 return 1;
 341
 342         return 0;
 343 }
 344 #endif
 345
 346 #ifdef CONFIG_ACPI_NUMA
 347 static int __init
 348 dmar_parse_one_rhsa(struct acpi_dmar_header *header)
 349 {
 350         struct acpi_dmar_rhsa *rhsa;
 351         struct dmar_drhd_unit *drhd;
 352
 353         rhsa = (struct acpi_dmar_rhsa *)header;
 354         for_each_drhd_unit(drhd) {
 355                 if (drhd->reg_base_addr == rhsa->base_address) {
 356                         int node = acpi_map_pxm_to_node(rhsa->proximity_domain);
 357
 358                         if (!node_online(node))
 359                                 node = -1;
 360                         drhd->iommu->node = node;
 361                         return 0;
 362                 }
 363         }
 364         WARN_TAINT(
 365                 1, TAINT_FIRMWARE_WORKAROUND,
 366                 "Your BIOS is broken; RHSA refers to non-existent DMAR unit at %llx\n"
 367                 "BIOS vendor: %s; Ver: %s; Product Version: %s\n",
 368                 drhd->reg_base_addr,
 369                 dmi_get_system_info(DMI_BIOS_VENDOR),
 370                 dmi_get_system_info(DMI_BIOS_VERSION),
 371                 dmi_get_system_info(DMI_PRODUCT_VERSION));
 372
 373         return 0;
 374 }
 375 #endif
 376
 377 static void __init
 378 dmar_table_print_dmar_entry(struct acpi_dmar_header *header)
 379 {
 380         struct acpi_dmar_hardware_unit *drhd;
 381         struct acpi_dmar_reserved_memory *rmrr;
 382         struct acpi_dmar_atsr *atsr;
 383         struct acpi_dmar_rhsa *rhsa;
 384
 385         switch (header->type) {
 386         case ACPI_DMAR_TYPE_HARDWARE_UNIT:
 387                 drhd = container_of(header, struct acpi_dmar_hardware_unit,
 388                                     header);
 389                 printk (KERN_INFO PREFIX
 390                         "DRHD base: %#016Lx flags: %#x\n",
 391                         (unsigned long long)drhd->address, drhd->flags);
 392                 break;
 393         case ACPI_DMAR_TYPE_RESERVED_MEMORY:
 394                 rmrr = container_of(header, struct acpi_dmar_reserved_memory,
 395                                     header);
 396                 printk (KERN_INFO PREFIX
 397                         "RMRR base: %#016Lx end: %#016Lx\n",
 398                         (unsigned long long)rmrr->base_address,
 399                         (unsigned long long)rmrr->end_address);
 400                 break;
 401         case ACPI_DMAR_TYPE_ATSR:
 402                 atsr = container_of(header, struct acpi_dmar_atsr, header);
 403                 printk(KERN_INFO PREFIX "ATSR flags: %#x\n", atsr->flags);
 404                 break;
 405         case ACPI_DMAR_HARDWARE_AFFINITY:
 406                 rhsa = container_of(header, struct acpi_dmar_rhsa, header);
 407                 printk(KERN_INFO PREFIX "RHSA base: %#016Lx proximity domain: %#x\n",
 408                        (unsigned long long)rhsa->base_address,
 409                        rhsa->proximity_domain);
 410                 break;
 411         }
 412 }
 413
 414 /**
 415  * dmar_table_detect - checks to see if the platform supports DMAR devices
 416  */
 417 static int __init dmar_table_detect(void)
 418 {
 419         acpi_status status = AE_OK;
 420
 421         /* if we could find DMAR table, then there are DMAR devices */
 422         status = acpi_get_table_with_size(ACPI_SIG_DMAR, 0,
 423                                 (struct acpi_table_header **)&dmar_tbl,
 424                                 &dmar_tbl_size);
 425
 426         if (ACPI_SUCCESS(status) && !dmar_tbl) {
 427                 printk (KERN_WARNING PREFIX "Unable to map DMAR\n");
 428                 status = AE_NOT_FOUND;
 429         }
 430
 431         return (ACPI_SUCCESS(status) ? 1 : 0);
 432 }
 433
 434 /**
 435  * parse_dmar_table - parses the DMA reporting table
 436  */
 437 static int __init
 438 parse_dmar_table(void)
 439 {
 440         struct acpi_table_dmar *dmar;
 441         struct acpi_dmar_header *entry_header;
 442         int ret = 0;
 443
 444         /*
 445          * Do it again, earlier dmar_tbl mapping could be mapped with
 446          * fixed map.
 447          */
 448         dmar_table_detect();
 449
 450         /*
 451          * ACPI tables may not be DMA protected by tboot, so use DMAR copy
 452          * SINIT saved in SinitMleData in TXT heap (which is DMA protected)
 453          */
 454         dmar_tbl = tboot_get_dmar_table(dmar_tbl);
 455
 456         dmar = (struct acpi_table_dmar *)dmar_tbl;
 457         if (!dmar)
 458                 return -ENODEV;
 459
 460         if (dmar->width < PAGE_SHIFT - 1) {
 461                 printk(KERN_WARNING PREFIX "Invalid DMAR haw\n");
 462                 return -EINVAL;
 463         }
 464
 465         printk (KERN_INFO PREFIX "Host address width %d\n",
 466                 dmar->width + 1);
 467
 468         entry_header = (struct acpi_dmar_header *)(dmar + 1);
 469         while (((unsigned long)entry_header) <
 470                         (((unsigned long)dmar) + dmar_tbl->length)) {
 471                 /* Avoid looping forever on bad ACPI tables */
 472                 if (entry_header->length == 0) {
 473                         printk(KERN_WARNING PREFIX
 474                                 "Invalid 0-length structure\n");
 475                         ret = -EINVAL;
 476                         break;
 477                 }
 478
 479                 dmar_table_print_dmar_entry(entry_header);
 480
 481                 switch (entry_header->type) {
 482                 case ACPI_DMAR_TYPE_HARDWARE_UNIT:
 483                         ret = dmar_parse_one_drhd(entry_header);
 484                         break;
 485                 case ACPI_DMAR_TYPE_RESERVED_MEMORY:
 486 #ifdef CONFIG_DMAR
 487                         ret = dmar_parse_one_rmrr(entry_header);
 488 #endif
 489                         break;
 490                 case ACPI_DMAR_TYPE_ATSR:
 491 #ifdef CONFIG_DMAR
 492                         ret = dmar_parse_one_atsr(entry_header);
 493 #endif
 494                         break;
 495                 case ACPI_DMAR_HARDWARE_AFFINITY:
 496 #ifdef CONFIG_ACPI_NUMA
 497                         ret = dmar_parse_one_rhsa(entry_header);
 498 #endif
 499                         break;
 500                 default:
 501                         printk(KERN_WARNING PREFIX
 502                                 "Unknown DMAR structure type %d\n",
 503                                 entry_header->type);
 504                         ret = 0; /* for forward compatibility */
 505                         break;
 506                 }
 507                 if (ret)
 508                         break;
 509
 510                 entry_header = ((void *)entry_header + entry_header->length);
 511         }
 512         return ret;
 513 }
 514
 515 static int dmar_pci_device_match(struct pci_dev *devices[], int cnt,
 516                           struct pci_dev *dev)
 517 {
 518         int index;
 519
 520         while (dev) {
 521                 for (index = 0; index < cnt; index++)
 522                         if (dev == devices[index])
 523                                 return 1;
 524
 525                 /* Check our parent */
 526                 dev = dev->bus->self;
 527         }
 528
 529         return 0;
 530 }
 531
 532 struct dmar_drhd_unit *
 533 dmar_find_matched_drhd_unit(struct pci_dev *dev)
 534 {
 535         struct dmar_drhd_unit *dmaru = NULL;
 536         struct acpi_dmar_hardware_unit *drhd;
 537
 538         dev = pci_physfn(dev);
 539
 540         list_for_each_entry(dmaru, &dmar_drhd_units, list) {
 541                 drhd = container_of(dmaru->hdr,
 542                                     struct acpi_dmar_hardware_unit,
 543                                     header);
 544
 545                 if (dmaru->include_all &&
 546                     drhd->segment == pci_domain_nr(dev->bus))
 547                         return dmaru;
 548
 549                 if (dmar_pci_device_match(dmaru->devices,
 550                                           dmaru->devices_cnt, dev))
 551                         return dmaru;
 552         }
 553
 554         return NULL;
 555 }
 556
 557 int __init dmar_dev_scope_init(void)
 558 {
 559         struct dmar_drhd_unit *drhd, *drhd_n;
 560         int ret = -ENODEV;
 561
 562         list_for_each_entry_safe(drhd, drhd_n, &dmar_drhd_units, list) {
 563                 ret = dmar_parse_dev(drhd);
 564                 if (ret)
 565                         return ret;
 566         }
 567
 568 #ifdef CONFIG_DMAR
 569         {
 570                 struct dmar_rmrr_unit *rmrr, *rmrr_n;
 571                 struct dmar_atsr_unit *atsr, *atsr_n;
 572
 573                 list_for_each_entry_safe(rmrr, rmrr_n, &dmar_rmrr_units, list) {
 574                         ret = rmrr_parse_dev(rmrr);
 575                         if (ret)
 576                                 return ret;
 577                 }
 578
 579                 list_for_each_entry_safe(atsr, atsr_n, &dmar_atsr_units, list) {
 580                         ret = atsr_parse_dev(atsr);
 581                         if (ret)
 582                                 return ret;
 583                 }
 584         }
 585 #endif
 586
 587         return ret;
 588 }
 589
 590
 591 int __init dmar_table_init(void)
 592 {
 593         static int dmar_table_initialized;
 594         int ret;
 595
 596         if (dmar_table_initialized)
 597                 return 0;
 598
 599         dmar_table_initialized = 1;
 600
 601         ret = parse_dmar_table();
 602         if (ret) {
 603                 if (ret != -ENODEV)
 604                         printk(KERN_INFO PREFIX "parse DMAR table failure.\n");
 605                 return ret;
 606         }
 607
 608         if (list_empty(&dmar_drhd_units)) {
 609                 printk(KERN_INFO PREFIX "No DMAR devices found\n");
 610                 return -ENODEV;
 611         }
 612
 613 #ifdef CONFIG_DMAR
 614         if (list_empty(&dmar_rmrr_units))
 615                 printk(KERN_INFO PREFIX "No RMRR found\n");
 616
 617         if (list_empty(&dmar_atsr_units))
 618                 printk(KERN_INFO PREFIX "No ATSR found\n");
 619 #endif
 620
 621         return 0;
 622 }
 623
 624 static void warn_invalid_dmar(u64 addr, const char *message)
 625 {
 626         WARN_TAINT_ONCE(
 627                 1, TAINT_FIRMWARE_WORKAROUND,
 628                 "Your BIOS is broken; DMAR reported at address %llx%s!\n"
 629                 "BIOS vendor: %s; Ver: %s; Product Version: %s\n",
 630                 addr, message,
 631                 dmi_get_system_info(DMI_BIOS_VENDOR),
 632                 dmi_get_system_info(DMI_BIOS_VERSION),
 633                 dmi_get_system_info(DMI_PRODUCT_VERSION));
 634 }
 635
 636 int __init check_zero_address(void)
 637 {
 638         struct acpi_table_dmar *dmar;
 639         struct acpi_dmar_header *entry_header;
 640         struct acpi_dmar_hardware_unit *drhd;
 641
 642         dmar = (struct acpi_table_dmar *)dmar_tbl;
 643         entry_header = (struct acpi_dmar_header *)(dmar + 1);
 644
 645         while (((unsigned long)entry_header) <
 646                         (((unsigned long)dmar) + dmar_tbl->length)) {
 647                 /* Avoid looping forever on bad ACPI tables */
 648                 if (entry_header->length == 0) {
 649                         printk(KERN_WARNING PREFIX
 650                                 "Invalid 0-length structure\n");
 651                         return 0;
 652                 }
 653
 654                 if (entry_header->type == ACPI_DMAR_TYPE_HARDWARE_UNIT) {
 655                         void __iomem *addr;
 656                         u64 cap, ecap;
 657
 658                         drhd = (void *)entry_header;
 659                         if (!drhd->address) {
 660                                 warn_invalid_dmar(0, "");
 661                                 goto failed;
 662                         }
 663
 664                         addr = early_ioremap(drhd->address, VTD_PAGE_SIZE);
 665                         if (!addr ) {
 666                                 printk("IOMMU: can't validate: %llx\n", drhd->address);
 667                                 goto failed;
 668                         }
 669                         cap = dmar_readq(addr + DMAR_CAP_REG);
 670                         ecap = dmar_readq(addr + DMAR_ECAP_REG);
 671                         early_iounmap(addr, VTD_PAGE_SIZE);
 672                         if (cap == (uint64_t)-1 && ecap == (uint64_t)-1) {
 673                                 warn_invalid_dmar(drhd->address,
 674                                                   " returns all ones");
 675                                 goto failed;
 676                         }
 677                 }
 678
 679                 entry_header = ((void *)entry_header + entry_header->length);
 680         }
 681         return 1;
 682
 683 failed:
 684 #ifdef CONFIG_DMAR
 685         dmar_disabled = 1;
 686 #endif
 687         return 0;
 688 }
 689
 690 void __init detect_intel_iommu(void)
 691 {
 692         int ret;
 693
 694         ret = dmar_table_detect();
 695         if (ret)
 696                 ret = check_zero_address();
 697         {
 698 #ifdef CONFIG_INTR_REMAP
 699                 struct acpi_table_dmar *dmar;
 700                 /*
 701                  * for now we will disable dma-remapping when interrupt
 702                  * remapping is enabled.
 703                  * When support for queued invalidation for IOTLB invalidation
 704                  * is added, we will not need this any more.
 705                  */
 706                 dmar = (struct acpi_table_dmar *) dmar_tbl;
 707                 if (ret && cpu_has_x2apic && dmar->flags & 0x1)
 708                         printk(KERN_INFO
 709                                "Queued invalidation will be enabled to support "
 710                                "x2apic and Intr-remapping.\n");
 711 #endif
 712 #ifdef CONFIG_DMAR
 713                 if (ret && !no_iommu && !iommu_detected && !dmar_disabled) {
 714                         iommu_detected = 1;
 715                         /* Make sure ACS will be enabled */
 716                         pci_request_acs();
 717                 }
 718 #endif
 719 #ifdef CONFIG_X86
 720                 if (ret)
 721                         x86_init.iommu.iommu_init = intel_iommu_init;
 722 #endif
 723         }
 724         early_acpi_os_unmap_memory(dmar_tbl, dmar_tbl_size);
 725         dmar_tbl = NULL;
 726 }
 727
 728
 729 int alloc_iommu(struct dmar_drhd_unit *drhd)
 730 {
 731         struct intel_iommu *iommu;
 732         int map_size;
 733         u32 ver;
 734         static int iommu_allocated = 0;
 735         int agaw = 0;
 736         int msagaw = 0;
 737
 738         if (!drhd->reg_base_addr) {
 739                 warn_invalid_dmar(0, "");
 740                 return -EINVAL;
 741         }
 742
 743         iommu = kzalloc(sizeof(*iommu), GFP_KERNEL);
 744         if (!iommu)
 745                 return -ENOMEM;
 746
 747         iommu->seq_id = iommu_allocated++;
 748         sprintf (iommu->name, "dmar%d", iommu->seq_id);
 749
 750         iommu->reg = ioremap(drhd->reg_base_addr, VTD_PAGE_SIZE);
 751         if (!iommu->reg) {
 752                 printk(KERN_ERR "IOMMU: can't map the region\n");
 753                 goto error;
 754         }
 755         iommu->cap = dmar_readq(iommu->reg + DMAR_CAP_REG);
 756         iommu->ecap = dmar_readq(iommu->reg + DMAR_ECAP_REG);
 757
 758         if (iommu->cap == (uint64_t)-1 && iommu->ecap == (uint64_t)-1) {
 759                 warn_invalid_dmar(drhd->reg_base_addr, " returns all ones");
 760                 goto err_unmap;
 761         }
 762
 763 #ifdef CONFIG_DMAR
 764         agaw = iommu_calculate_agaw(iommu);
 765         if (agaw < 0) {
 766                 printk(KERN_ERR
 767                        "Cannot get a valid agaw for iommu (seq_id = %d)\n",
 768                        iommu->seq_id);
 769                 goto err_unmap;
 770         }
 771         msagaw = iommu_calculate_max_sagaw(iommu);
 772         if (msagaw < 0) {
 773                 printk(KERN_ERR
 774                         "Cannot get a valid max agaw for iommu (seq_id = %d)\n",
 775                         iommu->seq_id);
 776                 goto err_unmap;
 777         }
 778 #endif
 779         iommu->agaw = agaw;
 780         iommu->msagaw = msagaw;
 781
 782         iommu->node = -1;
 783
 784         /* the registers might be more than one page */
 785         map_size = max_t(int, ecap_max_iotlb_offset(iommu->ecap),
 786                 cap_max_fault_reg_offset(iommu->cap));
 787         map_size = VTD_PAGE_ALIGN(map_size);
 788         if (map_size > VTD_PAGE_SIZE) {
 789                 iounmap(iommu->reg);
 790                 iommu->reg = ioremap(drhd->reg_base_addr, map_size);
 791                 if (!iommu->reg) {
 792                         printk(KERN_ERR "IOMMU: can't map the region\n");
 793                         goto error;
 794                 }
 795         }
 796
 797         ver = readl(iommu->reg + DMAR_VER_REG);
 798         pr_info("IOMMU %d: reg_base_addr %llx ver %d:%d cap %llx ecap %llx\n",
 799                 iommu->seq_id,
 800                 (unsigned long long)drhd->reg_base_addr,
 801                 DMAR_VER_MAJOR(ver), DMAR_VER_MINOR(ver),
 802                 (unsigned long long)iommu->cap,
 803                 (unsigned long long)iommu->ecap);
 804
 805         spin_lock_init(&iommu->register_lock);
 806
 807         drhd->iommu = iommu;
 808         return 0;
 809
 810  err_unmap:
 811         iounmap(iommu->reg);
 812  error:
 813         kfree(iommu);
 814         return -1;
 815 }
 816
 817 void free_iommu(struct intel_iommu *iommu)
 818 {
 819         if (!iommu)
 820                 return;
 821
 822 #ifdef CONFIG_DMAR
 823         free_dmar_iommu(iommu);
 824 #endif
 825
 826         if (iommu->reg)
 827                 iounmap(iommu->reg);
 828         kfree(iommu);
 829 }
 830
 831 /*
 832  * Reclaim all the submitted descriptors which have completed its work.
 833  */
 834 static inline void reclaim_free_desc(struct q_inval *qi)
 835 {
 836         while (qi->desc_status[qi->free_tail] == QI_DONE ||
 837                qi->desc_status[qi->free_tail] == QI_ABORT) {
 838                 qi->desc_status[qi->free_tail] = QI_FREE;
 839                 qi->free_tail = (qi->free_tail + 1) % QI_LENGTH;
 840                 qi->free_cnt++;
 841         }
 842 }
 843
 844 static int qi_check_fault(struct intel_iommu *iommu, int index)
 845 {
 846         u32 fault;
 847         int head, tail;
 848         struct q_inval *qi = iommu->qi;
 849         int wait_index = (index + 1) % QI_LENGTH;
 850
 851         if (qi->desc_status[wait_index] == QI_ABORT)
 852                 return -EAGAIN;
 853
 854         fault = readl(iommu->reg + DMAR_FSTS_REG);
 855
 856         /*
 857          * If IQE happens, the head points to the descriptor associated
 858          * with the error. No new descriptors are fetched until the IQE
 859          * is cleared.
 860          */
 861         if (fault & DMA_FSTS_IQE) {
 862                 head = readl(iommu->reg + DMAR_IQH_REG);
 863                 if ((head >> DMAR_IQ_SHIFT) == index) {
 864                         printk(KERN_ERR "VT-d detected invalid descriptor: "
 865                                 "low=%llx, high=%llx\n",
 866                                 (unsigned long long)qi->desc[index].low,
 867                                 (unsigned long long)qi->desc[index].high);
 868                         memcpy(&qi->desc[index], &qi->desc[wait_index],
 869                                         sizeof(struct qi_desc));
 870                         __iommu_flush_cache(iommu, &qi->desc[index],
 871                                         sizeof(struct qi_desc));
 872                         writel(DMA_FSTS_IQE, iommu->reg + DMAR_FSTS_REG);
 873                         return -EINVAL;
 874                 }
 875         }
 876
 877         /*
 878          * If ITE happens, all pending wait_desc commands are aborted.
 879          * No new descriptors are fetched until the ITE is cleared.
 880          */
 881         if (fault & DMA_FSTS_ITE) {
 882                 head = readl(iommu->reg + DMAR_IQH_REG);
 883                 head = ((head >> DMAR_IQ_SHIFT) - 1 + QI_LENGTH) % QI_LENGTH;
 884                 head |= 1;
 885                 tail = readl(iommu->reg + DMAR_IQT_REG);
 886                 tail = ((tail >> DMAR_IQ_SHIFT) - 1 + QI_LENGTH) % QI_LENGTH;
 887
 888                 writel(DMA_FSTS_ITE, iommu->reg + DMAR_FSTS_REG);
 889
 890                 do {
 891                         if (qi->desc_status[head] == QI_IN_USE)
 892                                 qi->desc_status[head] = QI_ABORT;
 893                         head = (head - 2 + QI_LENGTH) % QI_LENGTH;
 894                 } while (head != tail);
 895
 896                 if (qi->desc_status[wait_index] == QI_ABORT)
 897                         return -EAGAIN;
 898         }
 899
 900         if (fault & DMA_FSTS_ICE)
 901                 writel(DMA_FSTS_ICE, iommu->reg + DMAR_FSTS_REG);
 902
 903         return 0;
 904 }
 905
 906 /*
 907  * Submit the queued invalidation descriptor to the remapping
 908  * hardware unit and wait for its completion.
 909  */
 910 int qi_submit_sync(struct qi_desc *desc, struct intel_iommu *iommu)
 911 {
 912         int rc;
 913         struct q_inval *qi = iommu->qi;
 914         struct qi_desc *hw, wait_desc;
 915         int wait_index, index;
 916         unsigned long flags;
 917
 918         if (!qi)
 919                 return 0;
 920
 921         hw = qi->desc;
 922
 923 restart:
 924         rc = 0;
 925
 926         spin_lock_irqsave(&qi->q_lock, flags);
 927         while (qi->free_cnt < 3) {
 928                 spin_unlock_irqrestore(&qi->q_lock, flags);
 929                 cpu_relax();
 930                 spin_lock_irqsave(&qi->q_lock, flags);
 931         }
 932
 933         index = qi->free_head;
 934         wait_index = (index + 1) % QI_LENGTH;
 935
 936         qi->desc_status[index] = qi->desc_status[wait_index] = QI_IN_USE;
 937
 938         hw[index] = *desc;
 939
 940         wait_desc.low = QI_IWD_STATUS_DATA(QI_DONE) |
 941                         QI_IWD_STATUS_WRITE | QI_IWD_TYPE;
 942         wait_desc.high = virt_to_phys(&qi->desc_status[wait_index]);
 943
 944         hw[wait_index] = wait_desc;
 945
 946         __iommu_flush_cache(iommu, &hw[index], sizeof(struct qi_desc));
 947         __iommu_flush_cache(iommu, &hw[wait_index], sizeof(struct qi_desc));
 948
 949         qi->free_head = (qi->free_head + 2) % QI_LENGTH;
 950         qi->free_cnt -= 2;
 951
 952         /*
 953          * update the HW tail register indicating the presence of
 954          * new descriptors.
 955          */
 956         writel(qi->free_head << DMAR_IQ_SHIFT, iommu->reg + DMAR_IQT_REG);
 957
 958         while (qi->desc_status[wait_index] != QI_DONE) {
 959                 /*
 960                  * We will leave the interrupts disabled, to prevent interrupt
 961                  * context to queue another cmd while a cmd is already submitted
 962                  * and waiting for completion on this cpu. This is to avoid
 963                  * a deadlock where the interrupt context can wait indefinitely
 964                  * for free slots in the queue.
 965                  */
 966                 rc = qi_check_fault(iommu, index);
 967                 if (rc)
 968                         break;
 969
 970                 spin_unlock(&qi->q_lock);
 971                 cpu_relax();
 972                 spin_lock(&qi->q_lock);
 973         }
 974
 975         qi->desc_status[index] = QI_DONE;
 976
 977         reclaim_free_desc(qi);
 978         spin_unlock_irqrestore(&qi->q_lock, flags);
 979
 980         if (rc == -EAGAIN)
 981                 goto restart;
 982
 983         return rc;
 984 }
 985
 986 /*
 987  * Flush the global interrupt entry cache.
 988  */
 989 void qi_global_iec(struct intel_iommu *iommu)
 990 {
 991         struct qi_desc desc;
 992
 993         desc.low = QI_IEC_TYPE;
 994         desc.high = 0;
 995
 996         /* should never fail */
 997         qi_submit_sync(&desc, iommu);
 998 }
 999
1000 void qi_flush_context(struct intel_iommu *iommu, u16 did, u16 sid, u8 fm,
1001                       u64 type)
1002 {
1003         struct qi_desc desc;
1004
1005         desc.low = QI_CC_FM(fm) | QI_CC_SID(sid) | QI_CC_DID(did)
1006                         | QI_CC_GRAN(type) | QI_CC_TYPE;
1007         desc.high = 0;
1008
1009         qi_submit_sync(&desc, iommu);
1010 }
1011
1012 void qi_flush_iotlb(struct intel_iommu *iommu, u16 did, u64 addr,
1013                     unsigned int size_order, u64 type)
1014 {
1015         u8 dw = 0, dr = 0;
1016
1017         struct qi_desc desc;
1018         int ih = 0;
1019
1020         if (cap_write_drain(iommu->cap))
1021                 dw = 1;
1022
1023         if (cap_read_drain(iommu->cap))
1024                 dr = 1;
1025
1026         desc.low = QI_IOTLB_DID(did) | QI_IOTLB_DR(dr) | QI_IOTLB_DW(dw)
1027                 | QI_IOTLB_GRAN(type) | QI_IOTLB_TYPE;
1028         desc.high = QI_IOTLB_ADDR(addr) | QI_IOTLB_IH(ih)
1029                 | QI_IOTLB_AM(size_order);
1030
1031         qi_submit_sync(&desc, iommu);
1032 }
1033
1034 void qi_flush_dev_iotlb(struct intel_iommu *iommu, u16 sid, u16 qdep,
1035                         u64 addr, unsigned mask)
1036 {
1037         struct qi_desc desc;
1038
1039         if (mask) {
1040                 BUG_ON(addr & ((1 << (VTD_PAGE_SHIFT + mask)) - 1));
1041                 addr |= (1 << (VTD_PAGE_SHIFT + mask - 1)) - 1;
1042                 desc.high = QI_DEV_IOTLB_ADDR(addr) | QI_DEV_IOTLB_SIZE;
1043         } else
1044                 desc.high = QI_DEV_IOTLB_ADDR(addr);
1045
1046         if (qdep >= QI_DEV_IOTLB_MAX_INVS)
1047                 qdep = 0;
1048
1049         desc.low = QI_DEV_IOTLB_SID(sid) | QI_DEV_IOTLB_QDEP(qdep) |
1050                    QI_DIOTLB_TYPE;
1051
1052         qi_submit_sync(&desc, iommu);
1053 }
1054
1055 /*
1056  * Disable Queued Invalidation interface.
1057  */
1058 void dmar_disable_qi(struct intel_iommu *iommu)
1059 {
1060         unsigned long flags;
1061         u32 sts;
1062         cycles_t start_time = get_cycles();
1063
1064         if (!ecap_qis(iommu->ecap))
1065                 return;
1066
1067         spin_lock_irqsave(&iommu->register_lock, flags);
1068
1069         sts =  dmar_readq(iommu->reg + DMAR_GSTS_REG);
1070         if (!(sts & DMA_GSTS_QIES))
1071                 goto end;
1072
1073         /*
1074          * Give a chance to HW to complete the pending invalidation requests.
1075          */
1076         while ((readl(iommu->reg + DMAR_IQT_REG) !=
1077                 readl(iommu->reg + DMAR_IQH_REG)) &&
1078                 (DMAR_OPERATION_TIMEOUT > (get_cycles() - start_time)))
1079                 cpu_relax();
1080
1081         iommu->gcmd &= ~DMA_GCMD_QIE;
1082         writel(iommu->gcmd, iommu->reg + DMAR_GCMD_REG);
1083
1084         IOMMU_WAIT_OP(iommu, DMAR_GSTS_REG, readl,
1085                       !(sts & DMA_GSTS_QIES), sts);
1086 end:
1087         spin_unlock_irqrestore(&iommu->register_lock, flags);
1088 }
1089
1090 /*
1091  * Enable queued invalidation.
1092  */
1093 static void __dmar_enable_qi(struct intel_iommu *iommu)
1094 {
1095         u32 sts;
1096         unsigned long flags;
1097         struct q_inval *qi = iommu->qi;
1098
1099         qi->free_head = qi->free_tail = 0;
1100         qi->free_cnt = QI_LENGTH;
1101
1102         spin_lock_irqsave(&iommu->register_lock, flags);
1103
1104         /* write zero to the tail reg */
1105         writel(0, iommu->reg + DMAR_IQT_REG);
1106
1107         dmar_writeq(iommu->reg + DMAR_IQA_REG, virt_to_phys(qi->desc));
1108
1109         iommu->gcmd |= DMA_GCMD_QIE;
1110         writel(iommu->gcmd, iommu->reg + DMAR_GCMD_REG);
1111
1112         /* Make sure hardware complete it */
1113         IOMMU_WAIT_OP(iommu, DMAR_GSTS_REG, readl, (sts & DMA_GSTS_QIES), sts);
1114
1115         spin_unlock_irqrestore(&iommu->register_lock, flags);
1116 }
1117
1118 /*
1119  * Enable Queued Invalidation interface. This is a must to support
1120  * interrupt-remapping. Also used by DMA-remapping, which replaces
1121  * register based IOTLB invalidation.
1122  */
1123 int dmar_enable_qi(struct intel_iommu *iommu)
1124 {
1125         struct q_inval *qi;
1126         struct page *desc_page;
1127
1128         if (!ecap_qis(iommu->ecap))
1129                 return -ENOENT;
1130
1131         /*
1132          * queued invalidation is already setup and enabled.
1133          */
1134         if (iommu->qi)
1135                 return 0;
1136
1137         iommu->qi = kmalloc(sizeof(*qi), GFP_ATOMIC);
1138         if (!iommu->qi)
1139                 return -ENOMEM;
1140
1141         qi = iommu->qi;
1142
1143
1144         desc_page = alloc_pages_node(iommu->node, GFP_ATOMIC | __GFP_ZERO, 0);
1145         if (!desc_page) {
1146                 kfree(qi);
1147                 iommu->qi = 0;
1148                 return -ENOMEM;
1149         }
1150
1151         qi->desc = page_address(desc_page);
1152
1153         qi->desc_status = kmalloc(QI_LENGTH * sizeof(int), GFP_ATOMIC);
1154         if (!qi->desc_status) {
1155                 free_page((unsigned long) qi->desc);
1156                 kfree(qi);
1157                 iommu->qi = 0;
1158                 return -ENOMEM;
1159         }
1160
1161         qi->free_head = qi->free_tail = 0;
1162         qi->free_cnt = QI_LENGTH;
1163
1164         spin_lock_init(&qi->q_lock);
1165
1166         __dmar_enable_qi(iommu);
1167
1168         return 0;
1169 }
1170
1171 /* iommu interrupt handling. Most stuff are MSI-like. */
1172
1173 enum faulttype {
1174         DMA_REMAP,
1175         INTR_REMAP,
1176         UNKNOWN,
1177 };
1178
1179 static const char *dma_remap_fault_reasons[] =
1180 {
1181         "Software",
1182         "Present bit in root entry is clear",
1183         "Present bit in context entry is clear",
1184         "Invalid context entry",
1185         "Access beyond MGAW",
1186         "PTE Write access is not set",
1187         "PTE Read access is not set",
1188         "Next page table ptr is invalid",
1189         "Root table address invalid",
1190         "Context table ptr is invalid",
1191         "non-zero reserved fields in RTP",
1192         "non-zero reserved fields in CTP",
1193         "non-zero reserved fields in PTE",
1194 };
1195
1196 static const char *intr_remap_fault_reasons[] =
1197 {
1198         "Detected reserved fields in the decoded interrupt-remapped request",
1199         "Interrupt index exceeded the interrupt-remapping table size",
1200         "Present field in the IRTE entry is clear",
1201         "Error accessing interrupt-remapping table pointed by IRTA_REG",
1202         "Detected reserved fields in the IRTE entry",
1203         "Blocked a compatibility format interrupt request",
1204         "Blocked an interrupt request due to source-id verification failure",
1205 };
1206
1207 #define MAX_FAULT_REASON_IDX    (ARRAY_SIZE(fault_reason_strings) - 1)
1208
1209 const char *dmar_get_fault_reason(u8 fault_reason, int *fault_type)
1210 {
1211         if (fault_reason >= 0x20 && (fault_reason <= 0x20 +
1212                                      ARRAY_SIZE(intr_remap_fault_reasons))) {
1213                 *fault_type = INTR_REMAP;
1214                 return intr_remap_fault_reasons[fault_reason - 0x20];
1215         } else if (fault_reason < ARRAY_SIZE(dma_remap_fault_reasons)) {
1216                 *fault_type = DMA_REMAP;
1217                 return dma_remap_fault_reasons[fault_reason];
1218         } else {
1219                 *fault_type = UNKNOWN;
1220                 return "Unknown";
1221         }
1222 }
1223
1224 void dmar_msi_unmask(unsigned int irq)
1225 {
1226         struct intel_iommu *iommu = get_irq_data(irq);
1227         unsigned long flag;
1228
1229         /* unmask it */
1230         spin_lock_irqsave(&iommu->register_lock, flag);
1231         writel(0, iommu->reg + DMAR_FECTL_REG);
1232         /* Read a reg to force flush the post write */
1233         readl(iommu->reg + DMAR_FECTL_REG);
1234         spin_unlock_irqrestore(&iommu->register_lock, flag);
1235 }
1236
1237 void dmar_msi_mask(unsigned int irq)
1238 {
1239         unsigned long flag;
1240         struct intel_iommu *iommu = get_irq_data(irq);
1241
1242         /* mask it */
1243         spin_lock_irqsave(&iommu->register_lock, flag);
1244         writel(DMA_FECTL_IM, iommu->reg + DMAR_FECTL_REG);
1245         /* Read a reg to force flush the post write */
1246         readl(iommu->reg + DMAR_FECTL_REG);
1247         spin_unlock_irqrestore(&iommu->register_lock, flag);
1248 }
1249
1250 void dmar_msi_write(int irq, struct msi_msg *msg)
1251 {
1252         struct intel_iommu *iommu = get_irq_data(irq);
1253         unsigned long flag;
1254
1255         spin_lock_irqsave(&iommu->register_lock, flag);
1256         writel(msg->data, iommu->reg + DMAR_FEDATA_REG);
1257         writel(msg->address_lo, iommu->reg + DMAR_FEADDR_REG);
1258         writel(msg->address_hi, iommu->reg + DMAR_FEUADDR_REG);
1259         spin_unlock_irqrestore(&iommu->register_lock, flag);
1260 }
1261
1262 void dmar_msi_read(int irq, struct msi_msg *msg)
1263 {
1264         struct intel_iommu *iommu = get_irq_data(irq);
1265         unsigned long flag;
1266
1267         spin_lock_irqsave(&iommu->register_lock, flag);
1268         msg->data = readl(iommu->reg + DMAR_FEDATA_REG);
1269         msg->address_lo = readl(iommu->reg + DMAR_FEADDR_REG);
1270         msg->address_hi = readl(iommu->reg + DMAR_FEUADDR_REG);
1271         spin_unlock_irqrestore(&iommu->register_lock, flag);
1272 }
1273
1274 static int dmar_fault_do_one(struct intel_iommu *iommu, int type,
1275                 u8 fault_reason, u16 source_id, unsigned long long addr)
1276 {
1277         const char *reason;
1278         int fault_type;
1279
1280         reason = dmar_get_fault_reason(fault_reason, &fault_type);
1281
1282         if (fault_type == INTR_REMAP)
1283                 printk(KERN_ERR "INTR-REMAP: Request device [[%02x:%02x.%d] "
1284                        "fault index %llx\n"
1285                         "INTR-REMAP:[fault reason %02d] %s\n",
1286                         (source_id >> 8), PCI_SLOT(source_id & 0xFF),
1287                         PCI_FUNC(source_id & 0xFF), addr >> 48,
1288                         fault_reason, reason);
1289         else
1290                 printk(KERN_ERR
1291                        "DMAR:[%s] Request device [%02x:%02x.%d] "
1292                        "fault addr %llx \n"
1293                        "DMAR:[fault reason %02d] %s\n",
1294                        (type ? "DMA Read" : "DMA Write"),
1295                        (source_id >> 8), PCI_SLOT(source_id & 0xFF),
1296                        PCI_FUNC(source_id & 0xFF), addr, fault_reason, reason);
1297         return 0;
1298 }
1299
1300 #define PRIMARY_FAULT_REG_LEN (16)
1301 irqreturn_t dmar_fault(int irq, void *dev_id)
1302 {
1303         struct intel_iommu *iommu = dev_id;
1304         int reg, fault_index;
1305         u32 fault_status;
1306         unsigned long flag;
1307
1308         spin_lock_irqsave(&iommu->register_lock, flag);
1309         fault_status = readl(iommu->reg + DMAR_FSTS_REG);
1310         if (fault_status)
1311                 printk(KERN_ERR "DRHD: handling fault status reg %x\n",
1312                        fault_status);
1313
1314         /* TBD: ignore advanced fault log currently */
1315         if (!(fault_status & DMA_FSTS_PPF))
1316                 goto clear_rest;
1317
1318         fault_index = dma_fsts_fault_record_index(fault_status);
1319         reg = cap_fault_reg_offset(iommu->cap);
1320         while (1) {
1321                 u8 fault_reason;
1322                 u16 source_id;
1323                 u64 guest_addr;
1324                 int type;
1325                 u32 data;
1326
1327                 /* highest 32 bits */
1328                 data = readl(iommu->reg + reg +
1329                                 fault_index * PRIMARY_FAULT_REG_LEN + 12);
1330                 if (!(data & DMA_FRCD_F))
1331                         break;
1332
1333                 fault_reason = dma_frcd_fault_reason(data);
1334                 type = dma_frcd_type(data);
1335
1336                 data = readl(iommu->reg + reg +
1337                                 fault_index * PRIMARY_FAULT_REG_LEN + 8);
1338                 source_id = dma_frcd_source_id(data);
1339
1340                 guest_addr = dmar_readq(iommu->reg + reg +
1341                                 fault_index * PRIMARY_FAULT_REG_LEN);
1342                 guest_addr = dma_frcd_page_addr(guest_addr);
1343                 /* clear the fault */
1344                 writel(DMA_FRCD_F, iommu->reg + reg +
1345                         fault_index * PRIMARY_FAULT_REG_LEN + 12);
1346
1347                 spin_unlock_irqrestore(&iommu->register_lock, flag);
1348
1349                 dmar_fault_do_one(iommu, type, fault_reason,
1350                                 source_id, guest_addr);
1351
1352                 fault_index++;
1353                 if (fault_index >= cap_num_fault_regs(iommu->cap))
1354                         fault_index = 0;
1355                 spin_lock_irqsave(&iommu->register_lock, flag);
1356         }
1357 clear_rest:
1358         /* clear all the other faults */
1359         fault_status = readl(iommu->reg + DMAR_FSTS_REG);
1360         writel(fault_status, iommu->reg + DMAR_FSTS_REG);
1361
1362         spin_unlock_irqrestore(&iommu->register_lock, flag);
1363         return IRQ_HANDLED;
1364 }
1365
1366 int dmar_set_interrupt(struct intel_iommu *iommu)
1367 {
1368         int irq, ret;
1369
1370         /*
1371          * Check if the fault interrupt is already initialized.
1372          */
1373         if (iommu->irq)
1374                 return 0;
1375
1376         irq = create_irq();
1377         if (!irq) {
1378                 printk(KERN_ERR "IOMMU: no free vectors\n");
1379                 return -EINVAL;
1380         }
1381
1382         set_irq_data(irq, iommu);
1383         iommu->irq = irq;
1384
1385         ret = arch_setup_dmar_msi(irq);
1386         if (ret) {
1387                 set_irq_data(irq, NULL);
1388                 iommu->irq = 0;
1389                 destroy_irq(irq);
1390                 return ret;
1391         }
1392
1393         ret = request_irq(irq, dmar_fault, 0, iommu->name, iommu);
1394         if (ret)
1395                 printk(KERN_ERR "IOMMU: can't request irq\n");
1396         return ret;
1397 }
1398
1399 int __init enable_drhd_fault_handling(void)
1400 {
1401         struct dmar_drhd_unit *drhd;
1402
1403         /*
1404          * Enable fault control interrupt.
1405          */
1406         for_each_drhd_unit(drhd) {
1407                 int ret;
1408                 struct intel_iommu *iommu = drhd->iommu;
1409                 ret = dmar_set_interrupt(iommu);
1410
1411                 if (ret) {
1412                         printk(KERN_ERR "DRHD %Lx: failed to enable fault, "
1413                                " interrupt, ret %d\n",
1414                                (unsigned long long)drhd->reg_base_addr, ret);
1415                         return -1;
1416                 }
1417         }
1418
1419         return 0;
1420 }
1421
1422 /*
1423  * Re-enable Queued Invalidation interface.
1424  */
1425 int dmar_reenable_qi(struct intel_iommu *iommu)
1426 {
1427         if (!ecap_qis(iommu->ecap))
1428                 return -ENOENT;
1429
1430         if (!iommu->qi)
1431                 return -ENOENT;
1432
1433         /*
1434          * First disable queued invalidation.
1435          */
1436         dmar_disable_qi(iommu);
1437         /*
1438          * Then enable queued invalidation again. Since there is no pending
1439          * invalidation requests now, it's safe to re-enable queued
1440          * invalidation.
1441          */
1442         __dmar_enable_qi(iommu);
1443
1444         return 0;
1445 }
1446
1447 /*
1448  * Check interrupt remapping support in DMAR table description.
1449  */
1450 int __init dmar_ir_support(void)
1451 {
1452         struct acpi_table_dmar *dmar;
1453         dmar = (struct acpi_table_dmar *)dmar_tbl;
1454         if (!dmar)
1455                 return 0;
1456         return dmar->flags & 0x1;
1457 }