Merge branch 'master' of git://git2.kernel.org/pub/scm/linux/kernel/git/torvalds...
[linux-2.6/linux-2.6-openrd.git] / drivers / pci / dmar.c
blob83aae47475940b383fe89c5fd3cf30c7c7cb3ec4
1 /*
2 * Copyright (c) 2006, Intel Corporation.
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.
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.
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.
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>
22 * This file implements early detection/parsing of Remapping Devices
23 * reported to OS through BIOS via DMA remapping reporting (DMAR) ACPI
24 * tables.
26 * These routines are used by both DMA-remapping and Interrupt-remapping
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>
39 #define PREFIX "DMAR: "
41 /* No locks are needed as DMA remapping hardware unit
42 * list is constructed at boot time and hotplug of
43 * these units are not supported by the architecture.
45 LIST_HEAD(dmar_drhd_units);
47 static struct acpi_table_header * __initdata dmar_tbl;
48 static acpi_size dmar_tbl_size;
50 static void __init dmar_register_drhd_unit(struct dmar_drhd_unit *drhd)
53 * add INCLUDE_ALL at the tail, so scan the list will find it at
54 * the very end.
56 if (drhd->include_all)
57 list_add_tail(&drhd->list, &dmar_drhd_units);
58 else
59 list_add(&drhd->list, &dmar_drhd_units);
62 static int __init dmar_parse_one_dev_scope(struct acpi_dmar_device_scope *scope,
63 struct pci_dev **dev, u16 segment)
65 struct pci_bus *bus;
66 struct pci_dev *pdev = NULL;
67 struct acpi_dmar_pci_path *path;
68 int count;
70 bus = pci_find_bus(segment, scope->bus);
71 path = (struct acpi_dmar_pci_path *)(scope + 1);
72 count = (scope->length - sizeof(struct acpi_dmar_device_scope))
73 / sizeof(struct acpi_dmar_pci_path);
75 while (count) {
76 if (pdev)
77 pci_dev_put(pdev);
79 * Some BIOSes list non-exist devices in DMAR table, just
80 * ignore it
82 if (!bus) {
83 printk(KERN_WARNING
84 PREFIX "Device scope bus [%d] not found\n",
85 scope->bus);
86 break;
88 pdev = pci_get_slot(bus, PCI_DEVFN(path->dev, path->fn));
89 if (!pdev) {
90 printk(KERN_WARNING PREFIX
91 "Device scope device [%04x:%02x:%02x.%02x] not found\n",
92 segment, bus->number, path->dev, path->fn);
93 break;
95 path ++;
96 count --;
97 bus = pdev->subordinate;
99 if (!pdev) {
100 printk(KERN_WARNING PREFIX
101 "Device scope device [%04x:%02x:%02x.%02x] not found\n",
102 segment, scope->bus, path->dev, path->fn);
103 *dev = NULL;
104 return 0;
106 if ((scope->entry_type == ACPI_DMAR_SCOPE_TYPE_ENDPOINT && \
107 pdev->subordinate) || (scope->entry_type == \
108 ACPI_DMAR_SCOPE_TYPE_BRIDGE && !pdev->subordinate)) {
109 pci_dev_put(pdev);
110 printk(KERN_WARNING PREFIX
111 "Device scope type does not match for %s\n",
112 pci_name(pdev));
113 return -EINVAL;
115 *dev = pdev;
116 return 0;
119 static int __init dmar_parse_dev_scope(void *start, void *end, int *cnt,
120 struct pci_dev ***devices, u16 segment)
122 struct acpi_dmar_device_scope *scope;
123 void * tmp = start;
124 int index;
125 int ret;
127 *cnt = 0;
128 while (start < end) {
129 scope = start;
130 if (scope->entry_type == ACPI_DMAR_SCOPE_TYPE_ENDPOINT ||
131 scope->entry_type == ACPI_DMAR_SCOPE_TYPE_BRIDGE)
132 (*cnt)++;
133 else
134 printk(KERN_WARNING PREFIX
135 "Unsupported device scope\n");
136 start += scope->length;
138 if (*cnt == 0)
139 return 0;
141 *devices = kcalloc(*cnt, sizeof(struct pci_dev *), GFP_KERNEL);
142 if (!*devices)
143 return -ENOMEM;
145 start = tmp;
146 index = 0;
147 while (start < end) {
148 scope = start;
149 if (scope->entry_type == ACPI_DMAR_SCOPE_TYPE_ENDPOINT ||
150 scope->entry_type == ACPI_DMAR_SCOPE_TYPE_BRIDGE) {
151 ret = dmar_parse_one_dev_scope(scope,
152 &(*devices)[index], segment);
153 if (ret) {
154 kfree(*devices);
155 return ret;
157 index ++;
159 start += scope->length;
162 return 0;
166 * dmar_parse_one_drhd - parses exactly one DMA remapping hardware definition
167 * structure which uniquely represent one DMA remapping hardware unit
168 * present in the platform
170 static int __init
171 dmar_parse_one_drhd(struct acpi_dmar_header *header)
173 struct acpi_dmar_hardware_unit *drhd;
174 struct dmar_drhd_unit *dmaru;
175 int ret = 0;
177 drhd = (struct acpi_dmar_hardware_unit *)header;
178 dmaru = kzalloc(sizeof(*dmaru), GFP_KERNEL);
179 if (!dmaru)
180 return -ENOMEM;
182 dmaru->hdr = header;
183 dmaru->reg_base_addr = drhd->address;
184 dmaru->segment = drhd->segment;
185 dmaru->include_all = drhd->flags & 0x1; /* BIT0: INCLUDE_ALL */
187 ret = alloc_iommu(dmaru);
188 if (ret) {
189 kfree(dmaru);
190 return ret;
192 dmar_register_drhd_unit(dmaru);
193 return 0;
196 static int __init dmar_parse_dev(struct dmar_drhd_unit *dmaru)
198 struct acpi_dmar_hardware_unit *drhd;
199 int ret = 0;
201 drhd = (struct acpi_dmar_hardware_unit *) dmaru->hdr;
203 if (dmaru->include_all)
204 return 0;
206 ret = dmar_parse_dev_scope((void *)(drhd + 1),
207 ((void *)drhd) + drhd->header.length,
208 &dmaru->devices_cnt, &dmaru->devices,
209 drhd->segment);
210 if (ret) {
211 list_del(&dmaru->list);
212 kfree(dmaru);
214 return ret;
217 #ifdef CONFIG_DMAR
218 LIST_HEAD(dmar_rmrr_units);
220 static void __init dmar_register_rmrr_unit(struct dmar_rmrr_unit *rmrr)
222 list_add(&rmrr->list, &dmar_rmrr_units);
226 static int __init
227 dmar_parse_one_rmrr(struct acpi_dmar_header *header)
229 struct acpi_dmar_reserved_memory *rmrr;
230 struct dmar_rmrr_unit *rmrru;
232 rmrru = kzalloc(sizeof(*rmrru), GFP_KERNEL);
233 if (!rmrru)
234 return -ENOMEM;
236 rmrru->hdr = header;
237 rmrr = (struct acpi_dmar_reserved_memory *)header;
238 rmrru->base_address = rmrr->base_address;
239 rmrru->end_address = rmrr->end_address;
241 dmar_register_rmrr_unit(rmrru);
242 return 0;
245 static int __init
246 rmrr_parse_dev(struct dmar_rmrr_unit *rmrru)
248 struct acpi_dmar_reserved_memory *rmrr;
249 int ret;
251 rmrr = (struct acpi_dmar_reserved_memory *) rmrru->hdr;
252 ret = dmar_parse_dev_scope((void *)(rmrr + 1),
253 ((void *)rmrr) + rmrr->header.length,
254 &rmrru->devices_cnt, &rmrru->devices, rmrr->segment);
256 if (ret || (rmrru->devices_cnt == 0)) {
257 list_del(&rmrru->list);
258 kfree(rmrru);
260 return ret;
263 static LIST_HEAD(dmar_atsr_units);
265 static int __init dmar_parse_one_atsr(struct acpi_dmar_header *hdr)
267 struct acpi_dmar_atsr *atsr;
268 struct dmar_atsr_unit *atsru;
270 atsr = container_of(hdr, struct acpi_dmar_atsr, header);
271 atsru = kzalloc(sizeof(*atsru), GFP_KERNEL);
272 if (!atsru)
273 return -ENOMEM;
275 atsru->hdr = hdr;
276 atsru->include_all = atsr->flags & 0x1;
278 list_add(&atsru->list, &dmar_atsr_units);
280 return 0;
283 static int __init atsr_parse_dev(struct dmar_atsr_unit *atsru)
285 int rc;
286 struct acpi_dmar_atsr *atsr;
288 if (atsru->include_all)
289 return 0;
291 atsr = container_of(atsru->hdr, struct acpi_dmar_atsr, header);
292 rc = dmar_parse_dev_scope((void *)(atsr + 1),
293 (void *)atsr + atsr->header.length,
294 &atsru->devices_cnt, &atsru->devices,
295 atsr->segment);
296 if (rc || !atsru->devices_cnt) {
297 list_del(&atsru->list);
298 kfree(atsru);
301 return rc;
304 int dmar_find_matched_atsr_unit(struct pci_dev *dev)
306 int i;
307 struct pci_bus *bus;
308 struct acpi_dmar_atsr *atsr;
309 struct dmar_atsr_unit *atsru;
311 list_for_each_entry(atsru, &dmar_atsr_units, list) {
312 atsr = container_of(atsru->hdr, struct acpi_dmar_atsr, header);
313 if (atsr->segment == pci_domain_nr(dev->bus))
314 goto found;
317 return 0;
319 found:
320 for (bus = dev->bus; bus; bus = bus->parent) {
321 struct pci_dev *bridge = bus->self;
323 if (!bridge || !pci_is_pcie(bridge) ||
324 bridge->pcie_type == PCI_EXP_TYPE_PCI_BRIDGE)
325 return 0;
327 if (bridge->pcie_type == PCI_EXP_TYPE_ROOT_PORT) {
328 for (i = 0; i < atsru->devices_cnt; i++)
329 if (atsru->devices[i] == bridge)
330 return 1;
331 break;
335 if (atsru->include_all)
336 return 1;
338 return 0;
340 #endif
342 #ifdef CONFIG_ACPI_NUMA
343 static int __init
344 dmar_parse_one_rhsa(struct acpi_dmar_header *header)
346 struct acpi_dmar_rhsa *rhsa;
347 struct dmar_drhd_unit *drhd;
349 rhsa = (struct acpi_dmar_rhsa *)header;
350 for_each_drhd_unit(drhd) {
351 if (drhd->reg_base_addr == rhsa->base_address) {
352 int node = acpi_map_pxm_to_node(rhsa->proximity_domain);
354 if (!node_online(node))
355 node = -1;
356 drhd->iommu->node = node;
357 return 0;
360 WARN(1, "Your BIOS is broken; RHSA refers to non-existent DMAR unit at %llx\n"
361 "BIOS vendor: %s; Ver: %s; Product Version: %s\n",
362 drhd->reg_base_addr,
363 dmi_get_system_info(DMI_BIOS_VENDOR),
364 dmi_get_system_info(DMI_BIOS_VERSION),
365 dmi_get_system_info(DMI_PRODUCT_VERSION));
367 return 0;
369 #endif
371 static void __init
372 dmar_table_print_dmar_entry(struct acpi_dmar_header *header)
374 struct acpi_dmar_hardware_unit *drhd;
375 struct acpi_dmar_reserved_memory *rmrr;
376 struct acpi_dmar_atsr *atsr;
377 struct acpi_dmar_rhsa *rhsa;
379 switch (header->type) {
380 case ACPI_DMAR_TYPE_HARDWARE_UNIT:
381 drhd = container_of(header, struct acpi_dmar_hardware_unit,
382 header);
383 printk (KERN_INFO PREFIX
384 "DRHD base: %#016Lx flags: %#x\n",
385 (unsigned long long)drhd->address, drhd->flags);
386 break;
387 case ACPI_DMAR_TYPE_RESERVED_MEMORY:
388 rmrr = container_of(header, struct acpi_dmar_reserved_memory,
389 header);
390 printk (KERN_INFO PREFIX
391 "RMRR base: %#016Lx end: %#016Lx\n",
392 (unsigned long long)rmrr->base_address,
393 (unsigned long long)rmrr->end_address);
394 break;
395 case ACPI_DMAR_TYPE_ATSR:
396 atsr = container_of(header, struct acpi_dmar_atsr, header);
397 printk(KERN_INFO PREFIX "ATSR flags: %#x\n", atsr->flags);
398 break;
399 case ACPI_DMAR_HARDWARE_AFFINITY:
400 rhsa = container_of(header, struct acpi_dmar_rhsa, header);
401 printk(KERN_INFO PREFIX "RHSA base: %#016Lx proximity domain: %#x\n",
402 (unsigned long long)rhsa->base_address,
403 rhsa->proximity_domain);
404 break;
409 * dmar_table_detect - checks to see if the platform supports DMAR devices
411 static int __init dmar_table_detect(void)
413 acpi_status status = AE_OK;
415 /* if we could find DMAR table, then there are DMAR devices */
416 status = acpi_get_table_with_size(ACPI_SIG_DMAR, 0,
417 (struct acpi_table_header **)&dmar_tbl,
418 &dmar_tbl_size);
420 if (ACPI_SUCCESS(status) && !dmar_tbl) {
421 printk (KERN_WARNING PREFIX "Unable to map DMAR\n");
422 status = AE_NOT_FOUND;
425 return (ACPI_SUCCESS(status) ? 1 : 0);
429 * parse_dmar_table - parses the DMA reporting table
431 static int __init
432 parse_dmar_table(void)
434 struct acpi_table_dmar *dmar;
435 struct acpi_dmar_header *entry_header;
436 int ret = 0;
439 * Do it again, earlier dmar_tbl mapping could be mapped with
440 * fixed map.
442 dmar_table_detect();
445 * ACPI tables may not be DMA protected by tboot, so use DMAR copy
446 * SINIT saved in SinitMleData in TXT heap (which is DMA protected)
448 dmar_tbl = tboot_get_dmar_table(dmar_tbl);
450 dmar = (struct acpi_table_dmar *)dmar_tbl;
451 if (!dmar)
452 return -ENODEV;
454 if (dmar->width < PAGE_SHIFT - 1) {
455 printk(KERN_WARNING PREFIX "Invalid DMAR haw\n");
456 return -EINVAL;
459 printk (KERN_INFO PREFIX "Host address width %d\n",
460 dmar->width + 1);
462 entry_header = (struct acpi_dmar_header *)(dmar + 1);
463 while (((unsigned long)entry_header) <
464 (((unsigned long)dmar) + dmar_tbl->length)) {
465 /* Avoid looping forever on bad ACPI tables */
466 if (entry_header->length == 0) {
467 printk(KERN_WARNING PREFIX
468 "Invalid 0-length structure\n");
469 ret = -EINVAL;
470 break;
473 dmar_table_print_dmar_entry(entry_header);
475 switch (entry_header->type) {
476 case ACPI_DMAR_TYPE_HARDWARE_UNIT:
477 ret = dmar_parse_one_drhd(entry_header);
478 break;
479 case ACPI_DMAR_TYPE_RESERVED_MEMORY:
480 #ifdef CONFIG_DMAR
481 ret = dmar_parse_one_rmrr(entry_header);
482 #endif
483 break;
484 case ACPI_DMAR_TYPE_ATSR:
485 #ifdef CONFIG_DMAR
486 ret = dmar_parse_one_atsr(entry_header);
487 #endif
488 break;
489 case ACPI_DMAR_HARDWARE_AFFINITY:
490 #ifdef CONFIG_ACPI_NUMA
491 ret = dmar_parse_one_rhsa(entry_header);
492 #endif
493 break;
494 default:
495 printk(KERN_WARNING PREFIX
496 "Unknown DMAR structure type %d\n",
497 entry_header->type);
498 ret = 0; /* for forward compatibility */
499 break;
501 if (ret)
502 break;
504 entry_header = ((void *)entry_header + entry_header->length);
506 return ret;
509 int dmar_pci_device_match(struct pci_dev *devices[], int cnt,
510 struct pci_dev *dev)
512 int index;
514 while (dev) {
515 for (index = 0; index < cnt; index++)
516 if (dev == devices[index])
517 return 1;
519 /* Check our parent */
520 dev = dev->bus->self;
523 return 0;
526 struct dmar_drhd_unit *
527 dmar_find_matched_drhd_unit(struct pci_dev *dev)
529 struct dmar_drhd_unit *dmaru = NULL;
530 struct acpi_dmar_hardware_unit *drhd;
532 list_for_each_entry(dmaru, &dmar_drhd_units, list) {
533 drhd = container_of(dmaru->hdr,
534 struct acpi_dmar_hardware_unit,
535 header);
537 if (dmaru->include_all &&
538 drhd->segment == pci_domain_nr(dev->bus))
539 return dmaru;
541 if (dmar_pci_device_match(dmaru->devices,
542 dmaru->devices_cnt, dev))
543 return dmaru;
546 return NULL;
549 int __init dmar_dev_scope_init(void)
551 struct dmar_drhd_unit *drhd, *drhd_n;
552 int ret = -ENODEV;
554 list_for_each_entry_safe(drhd, drhd_n, &dmar_drhd_units, list) {
555 ret = dmar_parse_dev(drhd);
556 if (ret)
557 return ret;
560 #ifdef CONFIG_DMAR
562 struct dmar_rmrr_unit *rmrr, *rmrr_n;
563 struct dmar_atsr_unit *atsr, *atsr_n;
565 list_for_each_entry_safe(rmrr, rmrr_n, &dmar_rmrr_units, list) {
566 ret = rmrr_parse_dev(rmrr);
567 if (ret)
568 return ret;
571 list_for_each_entry_safe(atsr, atsr_n, &dmar_atsr_units, list) {
572 ret = atsr_parse_dev(atsr);
573 if (ret)
574 return ret;
577 #endif
579 return ret;
583 int __init dmar_table_init(void)
585 static int dmar_table_initialized;
586 int ret;
588 if (dmar_table_initialized)
589 return 0;
591 dmar_table_initialized = 1;
593 ret = parse_dmar_table();
594 if (ret) {
595 if (ret != -ENODEV)
596 printk(KERN_INFO PREFIX "parse DMAR table failure.\n");
597 return ret;
600 if (list_empty(&dmar_drhd_units)) {
601 printk(KERN_INFO PREFIX "No DMAR devices found\n");
602 return -ENODEV;
605 #ifdef CONFIG_DMAR
606 if (list_empty(&dmar_rmrr_units))
607 printk(KERN_INFO PREFIX "No RMRR found\n");
609 if (list_empty(&dmar_atsr_units))
610 printk(KERN_INFO PREFIX "No ATSR found\n");
611 #endif
613 return 0;
616 static int bios_warned;
618 int __init check_zero_address(void)
620 struct acpi_table_dmar *dmar;
621 struct acpi_dmar_header *entry_header;
622 struct acpi_dmar_hardware_unit *drhd;
624 dmar = (struct acpi_table_dmar *)dmar_tbl;
625 entry_header = (struct acpi_dmar_header *)(dmar + 1);
627 while (((unsigned long)entry_header) <
628 (((unsigned long)dmar) + dmar_tbl->length)) {
629 /* Avoid looping forever on bad ACPI tables */
630 if (entry_header->length == 0) {
631 printk(KERN_WARNING PREFIX
632 "Invalid 0-length structure\n");
633 return 0;
636 if (entry_header->type == ACPI_DMAR_TYPE_HARDWARE_UNIT) {
637 void __iomem *addr;
638 u64 cap, ecap;
640 drhd = (void *)entry_header;
641 if (!drhd->address) {
642 /* Promote an attitude of violence to a BIOS engineer today */
643 WARN(1, "Your BIOS is broken; DMAR reported at address zero!\n"
644 "BIOS vendor: %s; Ver: %s; Product Version: %s\n",
645 dmi_get_system_info(DMI_BIOS_VENDOR),
646 dmi_get_system_info(DMI_BIOS_VERSION),
647 dmi_get_system_info(DMI_PRODUCT_VERSION));
648 bios_warned = 1;
649 goto failed;
652 addr = early_ioremap(drhd->address, VTD_PAGE_SIZE);
653 if (!addr ) {
654 printk("IOMMU: can't validate: %llx\n", drhd->address);
655 goto failed;
657 cap = dmar_readq(addr + DMAR_CAP_REG);
658 ecap = dmar_readq(addr + DMAR_ECAP_REG);
659 early_iounmap(addr, VTD_PAGE_SIZE);
660 if (cap == (uint64_t)-1 && ecap == (uint64_t)-1) {
661 /* Promote an attitude of violence to a BIOS engineer today */
662 WARN(1, "Your BIOS is broken; DMAR reported at address %llx returns all ones!\n"
663 "BIOS vendor: %s; Ver: %s; Product Version: %s\n",
664 drhd->address,
665 dmi_get_system_info(DMI_BIOS_VENDOR),
666 dmi_get_system_info(DMI_BIOS_VERSION),
667 dmi_get_system_info(DMI_PRODUCT_VERSION));
668 bios_warned = 1;
669 goto failed;
673 entry_header = ((void *)entry_header + entry_header->length);
675 return 1;
677 failed:
678 #ifdef CONFIG_DMAR
679 dmar_disabled = 1;
680 #endif
681 return 0;
684 void __init detect_intel_iommu(void)
686 int ret;
688 ret = dmar_table_detect();
689 if (ret)
690 ret = check_zero_address();
692 #ifdef CONFIG_INTR_REMAP
693 struct acpi_table_dmar *dmar;
695 * for now we will disable dma-remapping when interrupt
696 * remapping is enabled.
697 * When support for queued invalidation for IOTLB invalidation
698 * is added, we will not need this any more.
700 dmar = (struct acpi_table_dmar *) dmar_tbl;
701 if (ret && cpu_has_x2apic && dmar->flags & 0x1)
702 printk(KERN_INFO
703 "Queued invalidation will be enabled to support "
704 "x2apic and Intr-remapping.\n");
705 #endif
706 #ifdef CONFIG_DMAR
707 if (ret && !no_iommu && !iommu_detected && !dmar_disabled) {
708 iommu_detected = 1;
709 /* Make sure ACS will be enabled */
710 pci_request_acs();
712 #endif
713 #ifdef CONFIG_X86
714 if (ret)
715 x86_init.iommu.iommu_init = intel_iommu_init;
716 #endif
718 early_acpi_os_unmap_memory(dmar_tbl, dmar_tbl_size);
719 dmar_tbl = NULL;
723 int alloc_iommu(struct dmar_drhd_unit *drhd)
725 struct intel_iommu *iommu;
726 int map_size;
727 u32 ver;
728 static int iommu_allocated = 0;
729 int agaw = 0;
730 int msagaw = 0;
732 if (!drhd->reg_base_addr) {
733 if (!bios_warned) {
734 WARN(1, "Your BIOS is broken; DMAR reported at address zero!\n"
735 "BIOS vendor: %s; Ver: %s; Product Version: %s\n",
736 dmi_get_system_info(DMI_BIOS_VENDOR),
737 dmi_get_system_info(DMI_BIOS_VERSION),
738 dmi_get_system_info(DMI_PRODUCT_VERSION));
739 bios_warned = 1;
741 return -EINVAL;
744 iommu = kzalloc(sizeof(*iommu), GFP_KERNEL);
745 if (!iommu)
746 return -ENOMEM;
748 iommu->seq_id = iommu_allocated++;
749 sprintf (iommu->name, "dmar%d", iommu->seq_id);
751 iommu->reg = ioremap(drhd->reg_base_addr, VTD_PAGE_SIZE);
752 if (!iommu->reg) {
753 printk(KERN_ERR "IOMMU: can't map the region\n");
754 goto error;
756 iommu->cap = dmar_readq(iommu->reg + DMAR_CAP_REG);
757 iommu->ecap = dmar_readq(iommu->reg + DMAR_ECAP_REG);
759 if (iommu->cap == (uint64_t)-1 && iommu->ecap == (uint64_t)-1) {
760 if (!bios_warned) {
761 /* Promote an attitude of violence to a BIOS engineer today */
762 WARN(1, "Your BIOS is broken; DMAR reported at address %llx returns all ones!\n"
763 "BIOS vendor: %s; Ver: %s; Product Version: %s\n",
764 drhd->reg_base_addr,
765 dmi_get_system_info(DMI_BIOS_VENDOR),
766 dmi_get_system_info(DMI_BIOS_VERSION),
767 dmi_get_system_info(DMI_PRODUCT_VERSION));
768 bios_warned = 1;
770 goto err_unmap;
773 #ifdef CONFIG_DMAR
774 agaw = iommu_calculate_agaw(iommu);
775 if (agaw < 0) {
776 printk(KERN_ERR
777 "Cannot get a valid agaw for iommu (seq_id = %d)\n",
778 iommu->seq_id);
779 goto err_unmap;
781 msagaw = iommu_calculate_max_sagaw(iommu);
782 if (msagaw < 0) {
783 printk(KERN_ERR
784 "Cannot get a valid max agaw for iommu (seq_id = %d)\n",
785 iommu->seq_id);
786 goto err_unmap;
788 #endif
789 iommu->agaw = agaw;
790 iommu->msagaw = msagaw;
792 iommu->node = -1;
794 /* the registers might be more than one page */
795 map_size = max_t(int, ecap_max_iotlb_offset(iommu->ecap),
796 cap_max_fault_reg_offset(iommu->cap));
797 map_size = VTD_PAGE_ALIGN(map_size);
798 if (map_size > VTD_PAGE_SIZE) {
799 iounmap(iommu->reg);
800 iommu->reg = ioremap(drhd->reg_base_addr, map_size);
801 if (!iommu->reg) {
802 printk(KERN_ERR "IOMMU: can't map the region\n");
803 goto error;
807 ver = readl(iommu->reg + DMAR_VER_REG);
808 pr_info("IOMMU %llx: ver %d:%d cap %llx ecap %llx\n",
809 (unsigned long long)drhd->reg_base_addr,
810 DMAR_VER_MAJOR(ver), DMAR_VER_MINOR(ver),
811 (unsigned long long)iommu->cap,
812 (unsigned long long)iommu->ecap);
814 spin_lock_init(&iommu->register_lock);
816 drhd->iommu = iommu;
817 return 0;
819 err_unmap:
820 iounmap(iommu->reg);
821 error:
822 kfree(iommu);
823 return -1;
826 void free_iommu(struct intel_iommu *iommu)
828 if (!iommu)
829 return;
831 #ifdef CONFIG_DMAR
832 free_dmar_iommu(iommu);
833 #endif
835 if (iommu->reg)
836 iounmap(iommu->reg);
837 kfree(iommu);
841 * Reclaim all the submitted descriptors which have completed its work.
843 static inline void reclaim_free_desc(struct q_inval *qi)
845 while (qi->desc_status[qi->free_tail] == QI_DONE ||
846 qi->desc_status[qi->free_tail] == QI_ABORT) {
847 qi->desc_status[qi->free_tail] = QI_FREE;
848 qi->free_tail = (qi->free_tail + 1) % QI_LENGTH;
849 qi->free_cnt++;
853 static int qi_check_fault(struct intel_iommu *iommu, int index)
855 u32 fault;
856 int head, tail;
857 struct q_inval *qi = iommu->qi;
858 int wait_index = (index + 1) % QI_LENGTH;
860 if (qi->desc_status[wait_index] == QI_ABORT)
861 return -EAGAIN;
863 fault = readl(iommu->reg + DMAR_FSTS_REG);
866 * If IQE happens, the head points to the descriptor associated
867 * with the error. No new descriptors are fetched until the IQE
868 * is cleared.
870 if (fault & DMA_FSTS_IQE) {
871 head = readl(iommu->reg + DMAR_IQH_REG);
872 if ((head >> DMAR_IQ_SHIFT) == index) {
873 printk(KERN_ERR "VT-d detected invalid descriptor: "
874 "low=%llx, high=%llx\n",
875 (unsigned long long)qi->desc[index].low,
876 (unsigned long long)qi->desc[index].high);
877 memcpy(&qi->desc[index], &qi->desc[wait_index],
878 sizeof(struct qi_desc));
879 __iommu_flush_cache(iommu, &qi->desc[index],
880 sizeof(struct qi_desc));
881 writel(DMA_FSTS_IQE, iommu->reg + DMAR_FSTS_REG);
882 return -EINVAL;
887 * If ITE happens, all pending wait_desc commands are aborted.
888 * No new descriptors are fetched until the ITE is cleared.
890 if (fault & DMA_FSTS_ITE) {
891 head = readl(iommu->reg + DMAR_IQH_REG);
892 head = ((head >> DMAR_IQ_SHIFT) - 1 + QI_LENGTH) % QI_LENGTH;
893 head |= 1;
894 tail = readl(iommu->reg + DMAR_IQT_REG);
895 tail = ((tail >> DMAR_IQ_SHIFT) - 1 + QI_LENGTH) % QI_LENGTH;
897 writel(DMA_FSTS_ITE, iommu->reg + DMAR_FSTS_REG);
899 do {
900 if (qi->desc_status[head] == QI_IN_USE)
901 qi->desc_status[head] = QI_ABORT;
902 head = (head - 2 + QI_LENGTH) % QI_LENGTH;
903 } while (head != tail);
905 if (qi->desc_status[wait_index] == QI_ABORT)
906 return -EAGAIN;
909 if (fault & DMA_FSTS_ICE)
910 writel(DMA_FSTS_ICE, iommu->reg + DMAR_FSTS_REG);
912 return 0;
916 * Submit the queued invalidation descriptor to the remapping
917 * hardware unit and wait for its completion.
919 int qi_submit_sync(struct qi_desc *desc, struct intel_iommu *iommu)
921 int rc;
922 struct q_inval *qi = iommu->qi;
923 struct qi_desc *hw, wait_desc;
924 int wait_index, index;
925 unsigned long flags;
927 if (!qi)
928 return 0;
930 hw = qi->desc;
932 restart:
933 rc = 0;
935 spin_lock_irqsave(&qi->q_lock, flags);
936 while (qi->free_cnt < 3) {
937 spin_unlock_irqrestore(&qi->q_lock, flags);
938 cpu_relax();
939 spin_lock_irqsave(&qi->q_lock, flags);
942 index = qi->free_head;
943 wait_index = (index + 1) % QI_LENGTH;
945 qi->desc_status[index] = qi->desc_status[wait_index] = QI_IN_USE;
947 hw[index] = *desc;
949 wait_desc.low = QI_IWD_STATUS_DATA(QI_DONE) |
950 QI_IWD_STATUS_WRITE | QI_IWD_TYPE;
951 wait_desc.high = virt_to_phys(&qi->desc_status[wait_index]);
953 hw[wait_index] = wait_desc;
955 __iommu_flush_cache(iommu, &hw[index], sizeof(struct qi_desc));
956 __iommu_flush_cache(iommu, &hw[wait_index], sizeof(struct qi_desc));
958 qi->free_head = (qi->free_head + 2) % QI_LENGTH;
959 qi->free_cnt -= 2;
962 * update the HW tail register indicating the presence of
963 * new descriptors.
965 writel(qi->free_head << DMAR_IQ_SHIFT, iommu->reg + DMAR_IQT_REG);
967 while (qi->desc_status[wait_index] != QI_DONE) {
969 * We will leave the interrupts disabled, to prevent interrupt
970 * context to queue another cmd while a cmd is already submitted
971 * and waiting for completion on this cpu. This is to avoid
972 * a deadlock where the interrupt context can wait indefinitely
973 * for free slots in the queue.
975 rc = qi_check_fault(iommu, index);
976 if (rc)
977 break;
979 spin_unlock(&qi->q_lock);
980 cpu_relax();
981 spin_lock(&qi->q_lock);
984 qi->desc_status[index] = QI_DONE;
986 reclaim_free_desc(qi);
987 spin_unlock_irqrestore(&qi->q_lock, flags);
989 if (rc == -EAGAIN)
990 goto restart;
992 return rc;
996 * Flush the global interrupt entry cache.
998 void qi_global_iec(struct intel_iommu *iommu)
1000 struct qi_desc desc;
1002 desc.low = QI_IEC_TYPE;
1003 desc.high = 0;
1005 /* should never fail */
1006 qi_submit_sync(&desc, iommu);
1009 void qi_flush_context(struct intel_iommu *iommu, u16 did, u16 sid, u8 fm,
1010 u64 type)
1012 struct qi_desc desc;
1014 desc.low = QI_CC_FM(fm) | QI_CC_SID(sid) | QI_CC_DID(did)
1015 | QI_CC_GRAN(type) | QI_CC_TYPE;
1016 desc.high = 0;
1018 qi_submit_sync(&desc, iommu);
1021 void qi_flush_iotlb(struct intel_iommu *iommu, u16 did, u64 addr,
1022 unsigned int size_order, u64 type)
1024 u8 dw = 0, dr = 0;
1026 struct qi_desc desc;
1027 int ih = 0;
1029 if (cap_write_drain(iommu->cap))
1030 dw = 1;
1032 if (cap_read_drain(iommu->cap))
1033 dr = 1;
1035 desc.low = QI_IOTLB_DID(did) | QI_IOTLB_DR(dr) | QI_IOTLB_DW(dw)
1036 | QI_IOTLB_GRAN(type) | QI_IOTLB_TYPE;
1037 desc.high = QI_IOTLB_ADDR(addr) | QI_IOTLB_IH(ih)
1038 | QI_IOTLB_AM(size_order);
1040 qi_submit_sync(&desc, iommu);
1043 void qi_flush_dev_iotlb(struct intel_iommu *iommu, u16 sid, u16 qdep,
1044 u64 addr, unsigned mask)
1046 struct qi_desc desc;
1048 if (mask) {
1049 BUG_ON(addr & ((1 << (VTD_PAGE_SHIFT + mask)) - 1));
1050 addr |= (1 << (VTD_PAGE_SHIFT + mask - 1)) - 1;
1051 desc.high = QI_DEV_IOTLB_ADDR(addr) | QI_DEV_IOTLB_SIZE;
1052 } else
1053 desc.high = QI_DEV_IOTLB_ADDR(addr);
1055 if (qdep >= QI_DEV_IOTLB_MAX_INVS)
1056 qdep = 0;
1058 desc.low = QI_DEV_IOTLB_SID(sid) | QI_DEV_IOTLB_QDEP(qdep) |
1059 QI_DIOTLB_TYPE;
1061 qi_submit_sync(&desc, iommu);
1065 * Disable Queued Invalidation interface.
1067 void dmar_disable_qi(struct intel_iommu *iommu)
1069 unsigned long flags;
1070 u32 sts;
1071 cycles_t start_time = get_cycles();
1073 if (!ecap_qis(iommu->ecap))
1074 return;
1076 spin_lock_irqsave(&iommu->register_lock, flags);
1078 sts = dmar_readq(iommu->reg + DMAR_GSTS_REG);
1079 if (!(sts & DMA_GSTS_QIES))
1080 goto end;
1083 * Give a chance to HW to complete the pending invalidation requests.
1085 while ((readl(iommu->reg + DMAR_IQT_REG) !=
1086 readl(iommu->reg + DMAR_IQH_REG)) &&
1087 (DMAR_OPERATION_TIMEOUT > (get_cycles() - start_time)))
1088 cpu_relax();
1090 iommu->gcmd &= ~DMA_GCMD_QIE;
1091 writel(iommu->gcmd, iommu->reg + DMAR_GCMD_REG);
1093 IOMMU_WAIT_OP(iommu, DMAR_GSTS_REG, readl,
1094 !(sts & DMA_GSTS_QIES), sts);
1095 end:
1096 spin_unlock_irqrestore(&iommu->register_lock, flags);
1100 * Enable queued invalidation.
1102 static void __dmar_enable_qi(struct intel_iommu *iommu)
1104 u32 sts;
1105 unsigned long flags;
1106 struct q_inval *qi = iommu->qi;
1108 qi->free_head = qi->free_tail = 0;
1109 qi->free_cnt = QI_LENGTH;
1111 spin_lock_irqsave(&iommu->register_lock, flags);
1113 /* write zero to the tail reg */
1114 writel(0, iommu->reg + DMAR_IQT_REG);
1116 dmar_writeq(iommu->reg + DMAR_IQA_REG, virt_to_phys(qi->desc));
1118 iommu->gcmd |= DMA_GCMD_QIE;
1119 writel(iommu->gcmd, iommu->reg + DMAR_GCMD_REG);
1121 /* Make sure hardware complete it */
1122 IOMMU_WAIT_OP(iommu, DMAR_GSTS_REG, readl, (sts & DMA_GSTS_QIES), sts);
1124 spin_unlock_irqrestore(&iommu->register_lock, flags);
1128 * Enable Queued Invalidation interface. This is a must to support
1129 * interrupt-remapping. Also used by DMA-remapping, which replaces
1130 * register based IOTLB invalidation.
1132 int dmar_enable_qi(struct intel_iommu *iommu)
1134 struct q_inval *qi;
1135 struct page *desc_page;
1137 if (!ecap_qis(iommu->ecap))
1138 return -ENOENT;
1141 * queued invalidation is already setup and enabled.
1143 if (iommu->qi)
1144 return 0;
1146 iommu->qi = kmalloc(sizeof(*qi), GFP_ATOMIC);
1147 if (!iommu->qi)
1148 return -ENOMEM;
1150 qi = iommu->qi;
1153 desc_page = alloc_pages_node(iommu->node, GFP_ATOMIC | __GFP_ZERO, 0);
1154 if (!desc_page) {
1155 kfree(qi);
1156 iommu->qi = 0;
1157 return -ENOMEM;
1160 qi->desc = page_address(desc_page);
1162 qi->desc_status = kmalloc(QI_LENGTH * sizeof(int), GFP_ATOMIC);
1163 if (!qi->desc_status) {
1164 free_page((unsigned long) qi->desc);
1165 kfree(qi);
1166 iommu->qi = 0;
1167 return -ENOMEM;
1170 qi->free_head = qi->free_tail = 0;
1171 qi->free_cnt = QI_LENGTH;
1173 spin_lock_init(&qi->q_lock);
1175 __dmar_enable_qi(iommu);
1177 return 0;
1180 /* iommu interrupt handling. Most stuff are MSI-like. */
1182 enum faulttype {
1183 DMA_REMAP,
1184 INTR_REMAP,
1185 UNKNOWN,
1188 static const char *dma_remap_fault_reasons[] =
1190 "Software",
1191 "Present bit in root entry is clear",
1192 "Present bit in context entry is clear",
1193 "Invalid context entry",
1194 "Access beyond MGAW",
1195 "PTE Write access is not set",
1196 "PTE Read access is not set",
1197 "Next page table ptr is invalid",
1198 "Root table address invalid",
1199 "Context table ptr is invalid",
1200 "non-zero reserved fields in RTP",
1201 "non-zero reserved fields in CTP",
1202 "non-zero reserved fields in PTE",
1205 static const char *intr_remap_fault_reasons[] =
1207 "Detected reserved fields in the decoded interrupt-remapped request",
1208 "Interrupt index exceeded the interrupt-remapping table size",
1209 "Present field in the IRTE entry is clear",
1210 "Error accessing interrupt-remapping table pointed by IRTA_REG",
1211 "Detected reserved fields in the IRTE entry",
1212 "Blocked a compatibility format interrupt request",
1213 "Blocked an interrupt request due to source-id verification failure",
1216 #define MAX_FAULT_REASON_IDX (ARRAY_SIZE(fault_reason_strings) - 1)
1218 const char *dmar_get_fault_reason(u8 fault_reason, int *fault_type)
1220 if (fault_reason >= 0x20 && (fault_reason <= 0x20 +
1221 ARRAY_SIZE(intr_remap_fault_reasons))) {
1222 *fault_type = INTR_REMAP;
1223 return intr_remap_fault_reasons[fault_reason - 0x20];
1224 } else if (fault_reason < ARRAY_SIZE(dma_remap_fault_reasons)) {
1225 *fault_type = DMA_REMAP;
1226 return dma_remap_fault_reasons[fault_reason];
1227 } else {
1228 *fault_type = UNKNOWN;
1229 return "Unknown";
1233 void dmar_msi_unmask(unsigned int irq)
1235 struct intel_iommu *iommu = get_irq_data(irq);
1236 unsigned long flag;
1238 /* unmask it */
1239 spin_lock_irqsave(&iommu->register_lock, flag);
1240 writel(0, iommu->reg + DMAR_FECTL_REG);
1241 /* Read a reg to force flush the post write */
1242 readl(iommu->reg + DMAR_FECTL_REG);
1243 spin_unlock_irqrestore(&iommu->register_lock, flag);
1246 void dmar_msi_mask(unsigned int irq)
1248 unsigned long flag;
1249 struct intel_iommu *iommu = get_irq_data(irq);
1251 /* mask it */
1252 spin_lock_irqsave(&iommu->register_lock, flag);
1253 writel(DMA_FECTL_IM, iommu->reg + DMAR_FECTL_REG);
1254 /* Read a reg to force flush the post write */
1255 readl(iommu->reg + DMAR_FECTL_REG);
1256 spin_unlock_irqrestore(&iommu->register_lock, flag);
1259 void dmar_msi_write(int irq, struct msi_msg *msg)
1261 struct intel_iommu *iommu = get_irq_data(irq);
1262 unsigned long flag;
1264 spin_lock_irqsave(&iommu->register_lock, flag);
1265 writel(msg->data, iommu->reg + DMAR_FEDATA_REG);
1266 writel(msg->address_lo, iommu->reg + DMAR_FEADDR_REG);
1267 writel(msg->address_hi, iommu->reg + DMAR_FEUADDR_REG);
1268 spin_unlock_irqrestore(&iommu->register_lock, flag);
1271 void dmar_msi_read(int irq, struct msi_msg *msg)
1273 struct intel_iommu *iommu = get_irq_data(irq);
1274 unsigned long flag;
1276 spin_lock_irqsave(&iommu->register_lock, flag);
1277 msg->data = readl(iommu->reg + DMAR_FEDATA_REG);
1278 msg->address_lo = readl(iommu->reg + DMAR_FEADDR_REG);
1279 msg->address_hi = readl(iommu->reg + DMAR_FEUADDR_REG);
1280 spin_unlock_irqrestore(&iommu->register_lock, flag);
1283 static int dmar_fault_do_one(struct intel_iommu *iommu, int type,
1284 u8 fault_reason, u16 source_id, unsigned long long addr)
1286 const char *reason;
1287 int fault_type;
1289 reason = dmar_get_fault_reason(fault_reason, &fault_type);
1291 if (fault_type == INTR_REMAP)
1292 printk(KERN_ERR "INTR-REMAP: Request device [[%02x:%02x.%d] "
1293 "fault index %llx\n"
1294 "INTR-REMAP:[fault reason %02d] %s\n",
1295 (source_id >> 8), PCI_SLOT(source_id & 0xFF),
1296 PCI_FUNC(source_id & 0xFF), addr >> 48,
1297 fault_reason, reason);
1298 else
1299 printk(KERN_ERR
1300 "DMAR:[%s] Request device [%02x:%02x.%d] "
1301 "fault addr %llx \n"
1302 "DMAR:[fault reason %02d] %s\n",
1303 (type ? "DMA Read" : "DMA Write"),
1304 (source_id >> 8), PCI_SLOT(source_id & 0xFF),
1305 PCI_FUNC(source_id & 0xFF), addr, fault_reason, reason);
1306 return 0;
1309 #define PRIMARY_FAULT_REG_LEN (16)
1310 irqreturn_t dmar_fault(int irq, void *dev_id)
1312 struct intel_iommu *iommu = dev_id;
1313 int reg, fault_index;
1314 u32 fault_status;
1315 unsigned long flag;
1317 spin_lock_irqsave(&iommu->register_lock, flag);
1318 fault_status = readl(iommu->reg + DMAR_FSTS_REG);
1319 if (fault_status)
1320 printk(KERN_ERR "DRHD: handling fault status reg %x\n",
1321 fault_status);
1323 /* TBD: ignore advanced fault log currently */
1324 if (!(fault_status & DMA_FSTS_PPF))
1325 goto clear_rest;
1327 fault_index = dma_fsts_fault_record_index(fault_status);
1328 reg = cap_fault_reg_offset(iommu->cap);
1329 while (1) {
1330 u8 fault_reason;
1331 u16 source_id;
1332 u64 guest_addr;
1333 int type;
1334 u32 data;
1336 /* highest 32 bits */
1337 data = readl(iommu->reg + reg +
1338 fault_index * PRIMARY_FAULT_REG_LEN + 12);
1339 if (!(data & DMA_FRCD_F))
1340 break;
1342 fault_reason = dma_frcd_fault_reason(data);
1343 type = dma_frcd_type(data);
1345 data = readl(iommu->reg + reg +
1346 fault_index * PRIMARY_FAULT_REG_LEN + 8);
1347 source_id = dma_frcd_source_id(data);
1349 guest_addr = dmar_readq(iommu->reg + reg +
1350 fault_index * PRIMARY_FAULT_REG_LEN);
1351 guest_addr = dma_frcd_page_addr(guest_addr);
1352 /* clear the fault */
1353 writel(DMA_FRCD_F, iommu->reg + reg +
1354 fault_index * PRIMARY_FAULT_REG_LEN + 12);
1356 spin_unlock_irqrestore(&iommu->register_lock, flag);
1358 dmar_fault_do_one(iommu, type, fault_reason,
1359 source_id, guest_addr);
1361 fault_index++;
1362 if (fault_index >= cap_num_fault_regs(iommu->cap))
1363 fault_index = 0;
1364 spin_lock_irqsave(&iommu->register_lock, flag);
1366 clear_rest:
1367 /* clear all the other faults */
1368 fault_status = readl(iommu->reg + DMAR_FSTS_REG);
1369 writel(fault_status, iommu->reg + DMAR_FSTS_REG);
1371 spin_unlock_irqrestore(&iommu->register_lock, flag);
1372 return IRQ_HANDLED;
1375 int dmar_set_interrupt(struct intel_iommu *iommu)
1377 int irq, ret;
1380 * Check if the fault interrupt is already initialized.
1382 if (iommu->irq)
1383 return 0;
1385 irq = create_irq();
1386 if (!irq) {
1387 printk(KERN_ERR "IOMMU: no free vectors\n");
1388 return -EINVAL;
1391 set_irq_data(irq, iommu);
1392 iommu->irq = irq;
1394 ret = arch_setup_dmar_msi(irq);
1395 if (ret) {
1396 set_irq_data(irq, NULL);
1397 iommu->irq = 0;
1398 destroy_irq(irq);
1399 return ret;
1402 ret = request_irq(irq, dmar_fault, 0, iommu->name, iommu);
1403 if (ret)
1404 printk(KERN_ERR "IOMMU: can't request irq\n");
1405 return ret;
1408 int __init enable_drhd_fault_handling(void)
1410 struct dmar_drhd_unit *drhd;
1413 * Enable fault control interrupt.
1415 for_each_drhd_unit(drhd) {
1416 int ret;
1417 struct intel_iommu *iommu = drhd->iommu;
1418 ret = dmar_set_interrupt(iommu);
1420 if (ret) {
1421 printk(KERN_ERR "DRHD %Lx: failed to enable fault, "
1422 " interrupt, ret %d\n",
1423 (unsigned long long)drhd->reg_base_addr, ret);
1424 return -1;
1428 return 0;
1432 * Re-enable Queued Invalidation interface.
1434 int dmar_reenable_qi(struct intel_iommu *iommu)
1436 if (!ecap_qis(iommu->ecap))
1437 return -ENOENT;
1439 if (!iommu->qi)
1440 return -ENOENT;
1443 * First disable queued invalidation.
1445 dmar_disable_qi(iommu);
1447 * Then enable queued invalidation again. Since there is no pending
1448 * invalidation requests now, it's safe to re-enable queued
1449 * invalidation.
1451 __dmar_enable_qi(iommu);
1453 return 0;
1457 * Check interrupt remapping support in DMAR table description.
1459 int dmar_ir_support(void)
1461 struct acpi_table_dmar *dmar;
1462 dmar = (struct acpi_table_dmar *)dmar_tbl;
1463 return dmar->flags & 0x1;