projects / deliverable / linux.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
iommu/vt-d: Make use of IOMMU sysfs support
[deliverable/linux.git] / drivers / iommu / 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 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt /* has to precede printk.h */
30
31 #include <linux/pci.h>
32 #include <linux/dmar.h>
33 #include <linux/iova.h>
34 #include <linux/intel-iommu.h>
35 #include <linux/timer.h>
36 #include <linux/irq.h>
37 #include <linux/interrupt.h>
38 #include <linux/tboot.h>
39 #include <linux/dmi.h>
40 #include <linux/slab.h>
41 #include <linux/iommu.h>
42 #include <asm/irq_remapping.h>
43 #include <asm/iommu_table.h>
44
45 #include "irq_remapping.h"
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 acpi_size dmar_tbl_size;
64 static int dmar_dev_scope_status = 1;
65
66 static int alloc_iommu(struct dmar_drhd_unit *drhd);
67 static void free_iommu(struct intel_iommu *iommu);
68
69 static void __init dmar_register_drhd_unit(struct dmar_drhd_unit *drhd)
70 {
71 /*
72 * add INCLUDE_ALL at the tail, so scan the list will find it at
73 * the very end.
74 */
75 if (drhd->include_all)
76 list_add_tail_rcu(&drhd->list, &dmar_drhd_units);
77 else
78 list_add_rcu(&drhd->list, &dmar_drhd_units);
79 }
80
81 void *dmar_alloc_dev_scope(void *start, void *end, int *cnt)
82 {
83 struct acpi_dmar_device_scope *scope;
84
85 *cnt = 0;
86 while (start < end) {
87 scope = start;
88 if (scope->entry_type == ACPI_DMAR_SCOPE_TYPE_ACPI ||
89 scope->entry_type == ACPI_DMAR_SCOPE_TYPE_ENDPOINT ||
90 scope->entry_type == ACPI_DMAR_SCOPE_TYPE_BRIDGE)
91 (*cnt)++;
92 else if (scope->entry_type != ACPI_DMAR_SCOPE_TYPE_IOAPIC &&
93 scope->entry_type != ACPI_DMAR_SCOPE_TYPE_HPET) {
94 pr_warn("Unsupported device scope\n");
95 }
96 start += scope->length;
97 }
98 if (*cnt == 0)
99 return NULL;
100
101 return kcalloc(*cnt, sizeof(struct dmar_dev_scope), GFP_KERNEL);
102 }
103
104 void dmar_free_dev_scope(struct dmar_dev_scope **devices, int *cnt)
105 {
106 int i;
107 struct device *tmp_dev;
108
109 if (*devices && *cnt) {
110 for_each_active_dev_scope(*devices, *cnt, i, tmp_dev)
111 put_device(tmp_dev);
112 kfree(*devices);
113 }
114
115 *devices = NULL;
116 *cnt = 0;
117 }
118
119 /* Optimize out kzalloc()/kfree() for normal cases */
120 static char dmar_pci_notify_info_buf[64];
121
122 static struct dmar_pci_notify_info *
123 dmar_alloc_pci_notify_info(struct pci_dev *dev, unsigned long event)
124 {
125 int level = 0;
126 size_t size;
127 struct pci_dev *tmp;
128 struct dmar_pci_notify_info *info;
129
130 BUG_ON(dev->is_virtfn);
131
132 /* Only generate path[] for device addition event */
133 if (event == BUS_NOTIFY_ADD_DEVICE)
134 for (tmp = dev; tmp; tmp = tmp->bus->self)
135 level++;
136
137 size = sizeof(*info) + level * sizeof(struct acpi_dmar_pci_path);
138 if (size <= sizeof(dmar_pci_notify_info_buf)) {
139 info = (struct dmar_pci_notify_info *)dmar_pci_notify_info_buf;
140 } else {
141 info = kzalloc(size, GFP_KERNEL);
142 if (!info) {
143 pr_warn("Out of memory when allocating notify_info "
144 "for %s.\n", pci_name(dev));
145 if (dmar_dev_scope_status == 0)
146 dmar_dev_scope_status = -ENOMEM;
147 return NULL;
148 }
149 }
150
151 info->event = event;
152 info->dev = dev;
153 info->seg = pci_domain_nr(dev->bus);
154 info->level = level;
155 if (event == BUS_NOTIFY_ADD_DEVICE) {
156 for (tmp = dev; tmp; tmp = tmp->bus->self) {
157 level--;
158 info->path[level].device = PCI_SLOT(tmp->devfn);
159 info->path[level].function = PCI_FUNC(tmp->devfn);
160 if (pci_is_root_bus(tmp->bus))
161 info->bus = tmp->bus->number;
162 }
163 }
164
165 return info;
166 }
167
168 static inline void dmar_free_pci_notify_info(struct dmar_pci_notify_info *info)
169 {
170 if ((void *)info != dmar_pci_notify_info_buf)
171 kfree(info);
172 }
173
174 static bool dmar_match_pci_path(struct dmar_pci_notify_info *info, int bus,
175 struct acpi_dmar_pci_path *path, int count)
176 {
177 int i;
178
179 if (info->bus != bus)
180 return false;
181 if (info->level != count)
182 return false;
183
184 for (i = 0; i < count; i++) {
185 if (path[i].device != info->path[i].device ||
186 path[i].function != info->path[i].function)
187 return false;
188 }
189
190 return true;
191 }
192
193 /* Return: > 0 if match found, 0 if no match found, < 0 if error happens */
194 int dmar_insert_dev_scope(struct dmar_pci_notify_info *info,
195 void *start, void*end, u16 segment,
196 struct dmar_dev_scope *devices,
197 int devices_cnt)
198 {
199 int i, level;
200 struct device *tmp, *dev = &info->dev->dev;
201 struct acpi_dmar_device_scope *scope;
202 struct acpi_dmar_pci_path *path;
203
204 if (segment != info->seg)
205 return 0;
206
207 for (; start < end; start += scope->length) {
208 scope = start;
209 if (scope->entry_type != ACPI_DMAR_SCOPE_TYPE_ENDPOINT &&
210 scope->entry_type != ACPI_DMAR_SCOPE_TYPE_BRIDGE)
211 continue;
212
213 path = (struct acpi_dmar_pci_path *)(scope + 1);
214 level = (scope->length - sizeof(*scope)) / sizeof(*path);
215 if (!dmar_match_pci_path(info, scope->bus, path, level))
216 continue;
217
218 if ((scope->entry_type == ACPI_DMAR_SCOPE_TYPE_ENDPOINT) ^
219 (info->dev->hdr_type == PCI_HEADER_TYPE_NORMAL)) {
220 pr_warn("Device scope type does not match for %s\n",
221 pci_name(info->dev));
222 return -EINVAL;
223 }
224
225 for_each_dev_scope(devices, devices_cnt, i, tmp)
226 if (tmp == NULL) {
227 devices[i].bus = info->dev->bus->number;
228 devices[i].devfn = info->dev->devfn;
229 rcu_assign_pointer(devices[i].dev,
230 get_device(dev));
231 return 1;
232 }
233 BUG_ON(i >= devices_cnt);
234 }
235
236 return 0;
237 }
238
239 int dmar_remove_dev_scope(struct dmar_pci_notify_info *info, u16 segment,
240 struct dmar_dev_scope *devices, int count)
241 {
242 int index;
243 struct device *tmp;
244
245 if (info->seg != segment)
246 return 0;
247
248 for_each_active_dev_scope(devices, count, index, tmp)
249 if (tmp == &info->dev->dev) {
250 rcu_assign_pointer(devices[index].dev, NULL);
251 synchronize_rcu();
252 put_device(tmp);
253 return 1;
254 }
255
256 return 0;
257 }
258
259 static int dmar_pci_bus_add_dev(struct dmar_pci_notify_info *info)
260 {
261 int ret = 0;
262 struct dmar_drhd_unit *dmaru;
263 struct acpi_dmar_hardware_unit *drhd;
264
265 for_each_drhd_unit(dmaru) {
266 if (dmaru->include_all)
267 continue;
268
269 drhd = container_of(dmaru->hdr,
270 struct acpi_dmar_hardware_unit, header);
271 ret = dmar_insert_dev_scope(info, (void *)(drhd + 1),
272 ((void *)drhd) + drhd->header.length,
273 dmaru->segment,
274 dmaru->devices, dmaru->devices_cnt);
275 if (ret != 0)
276 break;
277 }
278 if (ret >= 0)
279 ret = dmar_iommu_notify_scope_dev(info);
280 if (ret < 0 && dmar_dev_scope_status == 0)
281 dmar_dev_scope_status = ret;
282
283 return ret;
284 }
285
286 static void dmar_pci_bus_del_dev(struct dmar_pci_notify_info *info)
287 {
288 struct dmar_drhd_unit *dmaru;
289
290 for_each_drhd_unit(dmaru)
291 if (dmar_remove_dev_scope(info, dmaru->segment,
292 dmaru->devices, dmaru->devices_cnt))
293 break;
294 dmar_iommu_notify_scope_dev(info);
295 }
296
297 static int dmar_pci_bus_notifier(struct notifier_block *nb,
298 unsigned long action, void *data)
299 {
300 struct pci_dev *pdev = to_pci_dev(data);
301 struct dmar_pci_notify_info *info;
302
303 /* Only care about add/remove events for physical functions */
304 if (pdev->is_virtfn)
305 return NOTIFY_DONE;
306 if (action != BUS_NOTIFY_ADD_DEVICE && action != BUS_NOTIFY_DEL_DEVICE)
307 return NOTIFY_DONE;
308
309 info = dmar_alloc_pci_notify_info(pdev, action);
310 if (!info)
311 return NOTIFY_DONE;
312
313 down_write(&dmar_global_lock);
314 if (action == BUS_NOTIFY_ADD_DEVICE)
315 dmar_pci_bus_add_dev(info);
316 else if (action == BUS_NOTIFY_DEL_DEVICE)
317 dmar_pci_bus_del_dev(info);
318 up_write(&dmar_global_lock);
319
320 dmar_free_pci_notify_info(info);
321
322 return NOTIFY_OK;
323 }
324
325 static struct notifier_block dmar_pci_bus_nb = {
326 .notifier_call = dmar_pci_bus_notifier,
327 .priority = INT_MIN,
328 };
329
330 /**
331 * dmar_parse_one_drhd - parses exactly one DMA remapping hardware definition
332 * structure which uniquely represent one DMA remapping hardware unit
333 * present in the platform
334 */
335 static int __init
336 dmar_parse_one_drhd(struct acpi_dmar_header *header)
337 {
338 struct acpi_dmar_hardware_unit *drhd;
339 struct dmar_drhd_unit *dmaru;
340 int ret = 0;
341
342 drhd = (struct acpi_dmar_hardware_unit *)header;
343 dmaru = kzalloc(sizeof(*dmaru), GFP_KERNEL);
344 if (!dmaru)
345 return -ENOMEM;
346
347 dmaru->hdr = header;
348 dmaru->reg_base_addr = drhd->address;
349 dmaru->segment = drhd->segment;
350 dmaru->include_all = drhd->flags & 0x1; /* BIT0: INCLUDE_ALL */
351 dmaru->devices = dmar_alloc_dev_scope((void *)(drhd + 1),
352 ((void *)drhd) + drhd->header.length,
353 &dmaru->devices_cnt);
354 if (dmaru->devices_cnt && dmaru->devices == NULL) {
355 kfree(dmaru);
356 return -ENOMEM;
357 }
358
359 ret = alloc_iommu(dmaru);
360 if (ret) {
361 dmar_free_dev_scope(&dmaru->devices,
362 &dmaru->devices_cnt);
363 kfree(dmaru);
364 return ret;
365 }
366 dmar_register_drhd_unit(dmaru);
367 return 0;
368 }
369
370 static void dmar_free_drhd(struct dmar_drhd_unit *dmaru)
371 {
372 if (dmaru->devices && dmaru->devices_cnt)
373 dmar_free_dev_scope(&dmaru->devices, &dmaru->devices_cnt);
374 if (dmaru->iommu)
375 free_iommu(dmaru->iommu);
376 kfree(dmaru);
377 }
378
379 static int __init dmar_parse_one_andd(struct acpi_dmar_header *header)
380 {
381 struct acpi_dmar_andd *andd = (void *)header;
382
383 /* Check for NUL termination within the designated length */
384 if (strnlen(andd->object_name, header->length - 8) == header->length - 8) {
385 WARN_TAINT(1, TAINT_FIRMWARE_WORKAROUND,
386 "Your BIOS is broken; ANDD object name is not NUL-terminated\n"
387 "BIOS vendor: %s; Ver: %s; Product Version: %s\n",
388 dmi_get_system_info(DMI_BIOS_VENDOR),
389 dmi_get_system_info(DMI_BIOS_VERSION),
390 dmi_get_system_info(DMI_PRODUCT_VERSION));
391 return -EINVAL;
392 }
393 pr_info("ANDD device: %x name: %s\n", andd->device_number,
394 andd->object_name);
395
396 return 0;
397 }
398
399 #ifdef CONFIG_ACPI_NUMA
400 static int __init
401 dmar_parse_one_rhsa(struct acpi_dmar_header *header)
402 {
403 struct acpi_dmar_rhsa *rhsa;
404 struct dmar_drhd_unit *drhd;
405
406 rhsa = (struct acpi_dmar_rhsa *)header;
407 for_each_drhd_unit(drhd) {
408 if (drhd->reg_base_addr == rhsa->base_address) {
409 int node = acpi_map_pxm_to_node(rhsa->proximity_domain);
410
411 if (!node_online(node))
412 node = -1;
413 drhd->iommu->node = node;
414 return 0;
415 }
416 }
417 WARN_TAINT(
418 1, TAINT_FIRMWARE_WORKAROUND,
419 "Your BIOS is broken; RHSA refers to non-existent DMAR unit at %llx\n"
420 "BIOS vendor: %s; Ver: %s; Product Version: %s\n",
421 drhd->reg_base_addr,
422 dmi_get_system_info(DMI_BIOS_VENDOR),
423 dmi_get_system_info(DMI_BIOS_VERSION),
424 dmi_get_system_info(DMI_PRODUCT_VERSION));
425
426 return 0;
427 }
428 #endif
429
430 static void __init
431 dmar_table_print_dmar_entry(struct acpi_dmar_header *header)
432 {
433 struct acpi_dmar_hardware_unit *drhd;
434 struct acpi_dmar_reserved_memory *rmrr;
435 struct acpi_dmar_atsr *atsr;
436 struct acpi_dmar_rhsa *rhsa;
437
438 switch (header->type) {
439 case ACPI_DMAR_TYPE_HARDWARE_UNIT:
440 drhd = container_of(header, struct acpi_dmar_hardware_unit,
441 header);
442 pr_info("DRHD base: %#016Lx flags: %#x\n",
443 (unsigned long long)drhd->address, drhd->flags);
444 break;
445 case ACPI_DMAR_TYPE_RESERVED_MEMORY:
446 rmrr = container_of(header, struct acpi_dmar_reserved_memory,
447 header);
448 pr_info("RMRR base: %#016Lx end: %#016Lx\n",
449 (unsigned long long)rmrr->base_address,
450 (unsigned long long)rmrr->end_address);
451 break;
452 case ACPI_DMAR_TYPE_ATSR:
453 atsr = container_of(header, struct acpi_dmar_atsr, header);
454 pr_info("ATSR flags: %#x\n", atsr->flags);
455 break;
456 case ACPI_DMAR_HARDWARE_AFFINITY:
457 rhsa = container_of(header, struct acpi_dmar_rhsa, header);
458 pr_info("RHSA base: %#016Lx proximity domain: %#x\n",
459 (unsigned long long)rhsa->base_address,
460 rhsa->proximity_domain);
461 break;
462 case ACPI_DMAR_TYPE_ANDD:
463 /* We don't print this here because we need to sanity-check
464 it first. So print it in dmar_parse_one_andd() instead. */
465 break;
466 }
467 }
468
469 /**
470 * dmar_table_detect - checks to see if the platform supports DMAR devices
471 */
472 static int __init dmar_table_detect(void)
473 {
474 acpi_status status = AE_OK;
475
476 /* if we could find DMAR table, then there are DMAR devices */
477 status = acpi_get_table_with_size(ACPI_SIG_DMAR, 0,
478 (struct acpi_table_header **)&dmar_tbl,
479 &dmar_tbl_size);
480
481 if (ACPI_SUCCESS(status) && !dmar_tbl) {
482 pr_warn("Unable to map DMAR\n");
483 status = AE_NOT_FOUND;
484 }
485
486 return (ACPI_SUCCESS(status) ? 1 : 0);
487 }
488
489 /**
490 * parse_dmar_table - parses the DMA reporting table
491 */
492 static int __init
493 parse_dmar_table(void)
494 {
495 struct acpi_table_dmar *dmar;
496 struct acpi_dmar_header *entry_header;
497 int ret = 0;
498 int drhd_count = 0;
499
500 /*
501 * Do it again, earlier dmar_tbl mapping could be mapped with
502 * fixed map.
503 */
504 dmar_table_detect();
505
506 /*
507 * ACPI tables may not be DMA protected by tboot, so use DMAR copy
508 * SINIT saved in SinitMleData in TXT heap (which is DMA protected)
509 */
510 dmar_tbl = tboot_get_dmar_table(dmar_tbl);
511
512 dmar = (struct acpi_table_dmar *)dmar_tbl;
513 if (!dmar)
514 return -ENODEV;
515
516 if (dmar->width < PAGE_SHIFT - 1) {
517 pr_warn("Invalid DMAR haw\n");
518 return -EINVAL;
519 }
520
521 pr_info("Host address width %d\n", dmar->width + 1);
522
523 entry_header = (struct acpi_dmar_header *)(dmar + 1);
524 while (((unsigned long)entry_header) <
525 (((unsigned long)dmar) + dmar_tbl->length)) {
526 /* Avoid looping forever on bad ACPI tables */
527 if (entry_header->length == 0) {
528 pr_warn("Invalid 0-length structure\n");
529 ret = -EINVAL;
530 break;
531 }
532
533 dmar_table_print_dmar_entry(entry_header);
534
535 switch (entry_header->type) {
536 case ACPI_DMAR_TYPE_HARDWARE_UNIT:
537 drhd_count++;
538 ret = dmar_parse_one_drhd(entry_header);
539 break;
540 case ACPI_DMAR_TYPE_RESERVED_MEMORY:
541 ret = dmar_parse_one_rmrr(entry_header);
542 break;
543 case ACPI_DMAR_TYPE_ATSR:
544 ret = dmar_parse_one_atsr(entry_header);
545 break;
546 case ACPI_DMAR_HARDWARE_AFFINITY:
547 #ifdef CONFIG_ACPI_NUMA
548 ret = dmar_parse_one_rhsa(entry_header);
549 #endif
550 break;
551 case ACPI_DMAR_TYPE_ANDD:
552 ret = dmar_parse_one_andd(entry_header);
553 break;
554 default:
555 pr_warn("Unknown DMAR structure type %d\n",
556 entry_header->type);
557 ret = 0; /* for forward compatibility */
558 break;
559 }
560 if (ret)
561 break;
562
563 entry_header = ((void *)entry_header + entry_header->length);
564 }
565 if (drhd_count == 0)
566 pr_warn(FW_BUG "No DRHD structure found in DMAR table\n");
567 return ret;
568 }
569
570 static int dmar_pci_device_match(struct dmar_dev_scope devices[],
571 int cnt, struct pci_dev *dev)
572 {
573 int index;
574 struct device *tmp;
575
576 while (dev) {
577 for_each_active_dev_scope(devices, cnt, index, tmp)
578 if (dev_is_pci(tmp) && dev == to_pci_dev(tmp))
579 return 1;
580
581 /* Check our parent */
582 dev = dev->bus->self;
583 }
584
585 return 0;
586 }
587
588 struct dmar_drhd_unit *
589 dmar_find_matched_drhd_unit(struct pci_dev *dev)
590 {
591 struct dmar_drhd_unit *dmaru;
592 struct acpi_dmar_hardware_unit *drhd;
593
594 dev = pci_physfn(dev);
595
596 rcu_read_lock();
597 for_each_drhd_unit(dmaru) {
598 drhd = container_of(dmaru->hdr,
599 struct acpi_dmar_hardware_unit,
600 header);
601
602 if (dmaru->include_all &&
603 drhd->segment == pci_domain_nr(dev->bus))
604 goto out;
605
606 if (dmar_pci_device_match(dmaru->devices,
607 dmaru->devices_cnt, dev))
608 goto out;
609 }
610 dmaru = NULL;
611 out:
612 rcu_read_unlock();
613
614 return dmaru;
615 }
616
617 static void __init dmar_acpi_insert_dev_scope(u8 device_number,
618 struct acpi_device *adev)
619 {
620 struct dmar_drhd_unit *dmaru;
621 struct acpi_dmar_hardware_unit *drhd;
622 struct acpi_dmar_device_scope *scope;
623 struct device *tmp;
624 int i;
625 struct acpi_dmar_pci_path *path;
626
627 for_each_drhd_unit(dmaru) {
628 drhd = container_of(dmaru->hdr,
629 struct acpi_dmar_hardware_unit,
630 header);
631
632 for (scope = (void *)(drhd + 1);
633 (unsigned long)scope < ((unsigned long)drhd) + drhd->header.length;
634 scope = ((void *)scope) + scope->length) {
635 if (scope->entry_type != ACPI_DMAR_SCOPE_TYPE_ACPI)
636 continue;
637 if (scope->enumeration_id != device_number)
638 continue;
639
640 path = (void *)(scope + 1);
641 pr_info("ACPI device \"%s\" under DMAR at %llx as %02x:%02x.%d\n",
642 dev_name(&adev->dev), dmaru->reg_base_addr,
643 scope->bus, path->device, path->function);
644 for_each_dev_scope(dmaru->devices, dmaru->devices_cnt, i, tmp)
645 if (tmp == NULL) {
646 dmaru->devices[i].bus = scope->bus;
647 dmaru->devices[i].devfn = PCI_DEVFN(path->device,
648 path->function);
649 rcu_assign_pointer(dmaru->devices[i].dev,
650 get_device(&adev->dev));
651 return;
652 }
653 BUG_ON(i >= dmaru->devices_cnt);
654 }
655 }
656 pr_warn("No IOMMU scope found for ANDD enumeration ID %d (%s)\n",
657 device_number, dev_name(&adev->dev));
658 }
659
660 static int __init dmar_acpi_dev_scope_init(void)
661 {
662 struct acpi_dmar_andd *andd;
663
664 if (dmar_tbl == NULL)
665 return -ENODEV;
666
667 for (andd = (void *)dmar_tbl + sizeof(struct acpi_table_dmar);
668 ((unsigned long)andd) < ((unsigned long)dmar_tbl) + dmar_tbl->length;
669 andd = ((void *)andd) + andd->header.length) {
670 if (andd->header.type == ACPI_DMAR_TYPE_ANDD) {
671 acpi_handle h;
672 struct acpi_device *adev;
673
674 if (!ACPI_SUCCESS(acpi_get_handle(ACPI_ROOT_OBJECT,
675 andd->object_name,
676 &h))) {
677 pr_err("Failed to find handle for ACPI object %s\n",
678 andd->object_name);
679 continue;
680 }
681 acpi_bus_get_device(h, &adev);
682 if (!adev) {
683 pr_err("Failed to get device for ACPI object %s\n",
684 andd->object_name);
685 continue;
686 }
687 dmar_acpi_insert_dev_scope(andd->device_number, adev);
688 }
689 }
690 return 0;
691 }
692
693 int __init dmar_dev_scope_init(void)
694 {
695 struct pci_dev *dev = NULL;
696 struct dmar_pci_notify_info *info;
697
698 if (dmar_dev_scope_status != 1)
699 return dmar_dev_scope_status;
700
701 if (list_empty(&dmar_drhd_units)) {
702 dmar_dev_scope_status = -ENODEV;
703 } else {
704 dmar_dev_scope_status = 0;
705
706 dmar_acpi_dev_scope_init();
707
708 for_each_pci_dev(dev) {
709 if (dev->is_virtfn)
710 continue;
711
712 info = dmar_alloc_pci_notify_info(dev,
713 BUS_NOTIFY_ADD_DEVICE);
714 if (!info) {
715 return dmar_dev_scope_status;
716 } else {
717 dmar_pci_bus_add_dev(info);
718 dmar_free_pci_notify_info(info);
719 }
720 }
721
722 bus_register_notifier(&pci_bus_type, &dmar_pci_bus_nb);
723 }
724
725 return dmar_dev_scope_status;
726 }
727
728
729 int __init dmar_table_init(void)
730 {
731 static int dmar_table_initialized;
732 int ret;
733
734 if (dmar_table_initialized == 0) {
735 ret = parse_dmar_table();
736 if (ret < 0) {
737 if (ret != -ENODEV)
738 pr_info("parse DMAR table failure.\n");
739 } else if (list_empty(&dmar_drhd_units)) {
740 pr_info("No DMAR devices found\n");
741 ret = -ENODEV;
742 }
743
744 if (ret < 0)
745 dmar_table_initialized = ret;
746 else
747 dmar_table_initialized = 1;
748 }
749
750 return dmar_table_initialized < 0 ? dmar_table_initialized : 0;
751 }
752
753 static void warn_invalid_dmar(u64 addr, const char *message)
754 {
755 WARN_TAINT_ONCE(
756 1, TAINT_FIRMWARE_WORKAROUND,
757 "Your BIOS is broken; DMAR reported at address %llx%s!\n"
758 "BIOS vendor: %s; Ver: %s; Product Version: %s\n",
759 addr, message,
760 dmi_get_system_info(DMI_BIOS_VENDOR),
761 dmi_get_system_info(DMI_BIOS_VERSION),
762 dmi_get_system_info(DMI_PRODUCT_VERSION));
763 }
764
765 static int __init check_zero_address(void)
766 {
767 struct acpi_table_dmar *dmar;
768 struct acpi_dmar_header *entry_header;
769 struct acpi_dmar_hardware_unit *drhd;
770
771 dmar = (struct acpi_table_dmar *)dmar_tbl;
772 entry_header = (struct acpi_dmar_header *)(dmar + 1);
773
774 while (((unsigned long)entry_header) <
775 (((unsigned long)dmar) + dmar_tbl->length)) {
776 /* Avoid looping forever on bad ACPI tables */
777 if (entry_header->length == 0) {
778 pr_warn("Invalid 0-length structure\n");
779 return 0;
780 }
781
782 if (entry_header->type == ACPI_DMAR_TYPE_HARDWARE_UNIT) {
783 void __iomem *addr;
784 u64 cap, ecap;
785
786 drhd = (void *)entry_header;
787 if (!drhd->address) {
788 warn_invalid_dmar(0, "");
789 goto failed;
790 }
791
792 addr = early_ioremap(drhd->address, VTD_PAGE_SIZE);
793 if (!addr ) {
794 printk("IOMMU: can't validate: %llx\n", drhd->address);
795 goto failed;
796 }
797 cap = dmar_readq(addr + DMAR_CAP_REG);
798 ecap = dmar_readq(addr + DMAR_ECAP_REG);
799 early_iounmap(addr, VTD_PAGE_SIZE);
800 if (cap == (uint64_t)-1 && ecap == (uint64_t)-1) {
801 warn_invalid_dmar(drhd->address,
802 " returns all ones");
803 goto failed;
804 }
805 }
806
807 entry_header = ((void *)entry_header + entry_header->length);
808 }
809 return 1;
810
811 failed:
812 return 0;
813 }
814
815 int __init detect_intel_iommu(void)
816 {
817 int ret;
818
819 down_write(&dmar_global_lock);
820 ret = dmar_table_detect();
821 if (ret)
822 ret = check_zero_address();
823 {
824 if (ret && !no_iommu && !iommu_detected && !dmar_disabled) {
825 iommu_detected = 1;
826 /* Make sure ACS will be enabled */
827 pci_request_acs();
828 }
829
830 #ifdef CONFIG_X86
831 if (ret)
832 x86_init.iommu.iommu_init = intel_iommu_init;
833 #endif
834 }
835 early_acpi_os_unmap_memory((void __iomem *)dmar_tbl, dmar_tbl_size);
836 dmar_tbl = NULL;
837 up_write(&dmar_global_lock);
838
839 return ret ? 1 : -ENODEV;
840 }
841
842
843 static void unmap_iommu(struct intel_iommu *iommu)
844 {
845 iounmap(iommu->reg);
846 release_mem_region(iommu->reg_phys, iommu->reg_size);
847 }
848
849 /**
850 * map_iommu: map the iommu's registers
851 * @iommu: the iommu to map
852 * @phys_addr: the physical address of the base resgister
853 *
854 * Memory map the iommu's registers. Start w/ a single page, and
855 * possibly expand if that turns out to be insufficent.
856 */
857 static int map_iommu(struct intel_iommu *iommu, u64 phys_addr)
858 {
859 int map_size, err=0;
860
861 iommu->reg_phys = phys_addr;
862 iommu->reg_size = VTD_PAGE_SIZE;
863
864 if (!request_mem_region(iommu->reg_phys, iommu->reg_size, iommu->name)) {
865 pr_err("IOMMU: can't reserve memory\n");
866 err = -EBUSY;
867 goto out;
868 }
869
870 iommu->reg = ioremap(iommu->reg_phys, iommu->reg_size);
871 if (!iommu->reg) {
872 pr_err("IOMMU: can't map the region\n");
873 err = -ENOMEM;
874 goto release;
875 }
876
877 iommu->cap = dmar_readq(iommu->reg + DMAR_CAP_REG);
878 iommu->ecap = dmar_readq(iommu->reg + DMAR_ECAP_REG);
879
880 if (iommu->cap == (uint64_t)-1 && iommu->ecap == (uint64_t)-1) {
881 err = -EINVAL;
882 warn_invalid_dmar(phys_addr, " returns all ones");
883 goto unmap;
884 }
885
886 /* the registers might be more than one page */
887 map_size = max_t(int, ecap_max_iotlb_offset(iommu->ecap),
888 cap_max_fault_reg_offset(iommu->cap));
889 map_size = VTD_PAGE_ALIGN(map_size);
890 if (map_size > iommu->reg_size) {
891 iounmap(iommu->reg);
892 release_mem_region(iommu->reg_phys, iommu->reg_size);
893 iommu->reg_size = map_size;
894 if (!request_mem_region(iommu->reg_phys, iommu->reg_size,
895 iommu->name)) {
896 pr_err("IOMMU: can't reserve memory\n");
897 err = -EBUSY;
898 goto out;
899 }
900 iommu->reg = ioremap(iommu->reg_phys, iommu->reg_size);
901 if (!iommu->reg) {
902 pr_err("IOMMU: can't map the region\n");
903 err = -ENOMEM;
904 goto release;
905 }
906 }
907 err = 0;
908 goto out;
909
910 unmap:
911 iounmap(iommu->reg);
912 release:
913 release_mem_region(iommu->reg_phys, iommu->reg_size);
914 out:
915 return err;
916 }
917
918 static int alloc_iommu(struct dmar_drhd_unit *drhd)
919 {
920 struct intel_iommu *iommu;
921 u32 ver, sts;
922 static int iommu_allocated = 0;
923 int agaw = 0;
924 int msagaw = 0;
925 int err;
926
927 if (!drhd->reg_base_addr) {
928 warn_invalid_dmar(0, "");
929 return -EINVAL;
930 }
931
932 iommu = kzalloc(sizeof(*iommu), GFP_KERNEL);
933 if (!iommu)
934 return -ENOMEM;
935
936 iommu->seq_id = iommu_allocated++;
937 sprintf (iommu->name, "dmar%d", iommu->seq_id);
938
939 err = map_iommu(iommu, drhd->reg_base_addr);
940 if (err) {
941 pr_err("IOMMU: failed to map %s\n", iommu->name);
942 goto error;
943 }
944
945 err = -EINVAL;
946 agaw = iommu_calculate_agaw(iommu);
947 if (agaw < 0) {
948 pr_err("Cannot get a valid agaw for iommu (seq_id = %d)\n",
949 iommu->seq_id);
950 goto err_unmap;
951 }
952 msagaw = iommu_calculate_max_sagaw(iommu);
953 if (msagaw < 0) {
954 pr_err("Cannot get a valid max agaw for iommu (seq_id = %d)\n",
955 iommu->seq_id);
956 goto err_unmap;
957 }
958 iommu->agaw = agaw;
959 iommu->msagaw = msagaw;
960 iommu->segment = drhd->segment;
961
962 iommu->node = -1;
963
964 ver = readl(iommu->reg + DMAR_VER_REG);
965 pr_info("IOMMU %d: reg_base_addr %llx ver %d:%d cap %llx ecap %llx\n",
966 iommu->seq_id,
967 (unsigned long long)drhd->reg_base_addr,
968 DMAR_VER_MAJOR(ver), DMAR_VER_MINOR(ver),
969 (unsigned long long)iommu->cap,
970 (unsigned long long)iommu->ecap);
971
972 /* Reflect status in gcmd */
973 sts = readl(iommu->reg + DMAR_GSTS_REG);
974 if (sts & DMA_GSTS_IRES)
975 iommu->gcmd |= DMA_GCMD_IRE;
976 if (sts & DMA_GSTS_TES)
977 iommu->gcmd |= DMA_GCMD_TE;
978 if (sts & DMA_GSTS_QIES)
979 iommu->gcmd |= DMA_GCMD_QIE;
980
981 raw_spin_lock_init(&iommu->register_lock);
982
983 drhd->iommu = iommu;
984
985 if (intel_iommu_enabled)
986 iommu->iommu_dev = iommu_device_create(NULL, iommu,
987 intel_iommu_groups,
988 iommu->name);
989
990 return 0;
991
992 err_unmap:
993 unmap_iommu(iommu);
994 error:
995 kfree(iommu);
996 return err;
997 }
998
999 static void free_iommu(struct intel_iommu *iommu)
1000 {
1001 iommu_device_destroy(iommu->iommu_dev);
1002
1003 if (iommu->irq) {
1004 free_irq(iommu->irq, iommu);
1005 irq_set_handler_data(iommu->irq, NULL);
1006 dmar_free_hwirq(iommu->irq);
1007 }
1008
1009 if (iommu->qi) {
1010 free_page((unsigned long)iommu->qi->desc);
1011 kfree(iommu->qi->desc_status);
1012 kfree(iommu->qi);
1013 }
1014
1015 if (iommu->reg)
1016 unmap_iommu(iommu);
1017
1018 kfree(iommu);
1019 }
1020
1021 /*
1022 * Reclaim all the submitted descriptors which have completed its work.
1023 */
1024 static inline void reclaim_free_desc(struct q_inval *qi)
1025 {
1026 while (qi->desc_status[qi->free_tail] == QI_DONE ||
1027 qi->desc_status[qi->free_tail] == QI_ABORT) {
1028 qi->desc_status[qi->free_tail] = QI_FREE;
1029 qi->free_tail = (qi->free_tail + 1) % QI_LENGTH;
1030 qi->free_cnt++;
1031 }
1032 }
1033
1034 static int qi_check_fault(struct intel_iommu *iommu, int index)
1035 {
1036 u32 fault;
1037 int head, tail;
1038 struct q_inval *qi = iommu->qi;
1039 int wait_index = (index + 1) % QI_LENGTH;
1040
1041 if (qi->desc_status[wait_index] == QI_ABORT)
1042 return -EAGAIN;
1043
1044 fault = readl(iommu->reg + DMAR_FSTS_REG);
1045
1046 /*
1047 * If IQE happens, the head points to the descriptor associated
1048 * with the error. No new descriptors are fetched until the IQE
1049 * is cleared.
1050 */
1051 if (fault & DMA_FSTS_IQE) {
1052 head = readl(iommu->reg + DMAR_IQH_REG);
1053 if ((head >> DMAR_IQ_SHIFT) == index) {
1054 pr_err("VT-d detected invalid descriptor: "
1055 "low=%llx, high=%llx\n",
1056 (unsigned long long)qi->desc[index].low,
1057 (unsigned long long)qi->desc[index].high);
1058 memcpy(&qi->desc[index], &qi->desc[wait_index],
1059 sizeof(struct qi_desc));
1060 __iommu_flush_cache(iommu, &qi->desc[index],
1061 sizeof(struct qi_desc));
1062 writel(DMA_FSTS_IQE, iommu->reg + DMAR_FSTS_REG);
1063 return -EINVAL;
1064 }
1065 }
1066
1067 /*
1068 * If ITE happens, all pending wait_desc commands are aborted.
1069 * No new descriptors are fetched until the ITE is cleared.
1070 */
1071 if (fault & DMA_FSTS_ITE) {
1072 head = readl(iommu->reg + DMAR_IQH_REG);
1073 head = ((head >> DMAR_IQ_SHIFT) - 1 + QI_LENGTH) % QI_LENGTH;
1074 head |= 1;
1075 tail = readl(iommu->reg + DMAR_IQT_REG);
1076 tail = ((tail >> DMAR_IQ_SHIFT) - 1 + QI_LENGTH) % QI_LENGTH;
1077
1078 writel(DMA_FSTS_ITE, iommu->reg + DMAR_FSTS_REG);
1079
1080 do {
1081 if (qi->desc_status[head] == QI_IN_USE)
1082 qi->desc_status[head] = QI_ABORT;
1083 head = (head - 2 + QI_LENGTH) % QI_LENGTH;
1084 } while (head != tail);
1085
1086 if (qi->desc_status[wait_index] == QI_ABORT)
1087 return -EAGAIN;
1088 }
1089
1090 if (fault & DMA_FSTS_ICE)
1091 writel(DMA_FSTS_ICE, iommu->reg + DMAR_FSTS_REG);
1092
1093 return 0;
1094 }
1095
1096 /*
1097 * Submit the queued invalidation descriptor to the remapping
1098 * hardware unit and wait for its completion.
1099 */
1100 int qi_submit_sync(struct qi_desc *desc, struct intel_iommu *iommu)
1101 {
1102 int rc;
1103 struct q_inval *qi = iommu->qi;
1104 struct qi_desc *hw, wait_desc;
1105 int wait_index, index;
1106 unsigned long flags;
1107
1108 if (!qi)
1109 return 0;
1110
1111 hw = qi->desc;
1112
1113 restart:
1114 rc = 0;
1115
1116 raw_spin_lock_irqsave(&qi->q_lock, flags);
1117 while (qi->free_cnt < 3) {
1118 raw_spin_unlock_irqrestore(&qi->q_lock, flags);
1119 cpu_relax();
1120 raw_spin_lock_irqsave(&qi->q_lock, flags);
1121 }
1122
1123 index = qi->free_head;
1124 wait_index = (index + 1) % QI_LENGTH;
1125
1126 qi->desc_status[index] = qi->desc_status[wait_index] = QI_IN_USE;
1127
1128 hw[index] = *desc;
1129
1130 wait_desc.low = QI_IWD_STATUS_DATA(QI_DONE) |
1131 QI_IWD_STATUS_WRITE | QI_IWD_TYPE;
1132 wait_desc.high = virt_to_phys(&qi->desc_status[wait_index]);
1133
1134 hw[wait_index] = wait_desc;
1135
1136 __iommu_flush_cache(iommu, &hw[index], sizeof(struct qi_desc));
1137 __iommu_flush_cache(iommu, &hw[wait_index], sizeof(struct qi_desc));
1138
1139 qi->free_head = (qi->free_head + 2) % QI_LENGTH;
1140 qi->free_cnt -= 2;
1141
1142 /*
1143 * update the HW tail register indicating the presence of
1144 * new descriptors.
1145 */
1146 writel(qi->free_head << DMAR_IQ_SHIFT, iommu->reg + DMAR_IQT_REG);
1147
1148 while (qi->desc_status[wait_index] != QI_DONE) {
1149 /*
1150 * We will leave the interrupts disabled, to prevent interrupt
1151 * context to queue another cmd while a cmd is already submitted
1152 * and waiting for completion on this cpu. This is to avoid
1153 * a deadlock where the interrupt context can wait indefinitely
1154 * for free slots in the queue.
1155 */
1156 rc = qi_check_fault(iommu, index);
1157 if (rc)
1158 break;
1159
1160 raw_spin_unlock(&qi->q_lock);
1161 cpu_relax();
1162 raw_spin_lock(&qi->q_lock);
1163 }
1164
1165 qi->desc_status[index] = QI_DONE;
1166
1167 reclaim_free_desc(qi);
1168 raw_spin_unlock_irqrestore(&qi->q_lock, flags);
1169
1170 if (rc == -EAGAIN)
1171 goto restart;
1172
1173 return rc;
1174 }
1175
1176 /*
1177 * Flush the global interrupt entry cache.
1178 */
1179 void qi_global_iec(struct intel_iommu *iommu)
1180 {
1181 struct qi_desc desc;
1182
1183 desc.low = QI_IEC_TYPE;
1184 desc.high = 0;
1185
1186 /* should never fail */
1187 qi_submit_sync(&desc, iommu);
1188 }
1189
1190 void qi_flush_context(struct intel_iommu *iommu, u16 did, u16 sid, u8 fm,
1191 u64 type)
1192 {
1193 struct qi_desc desc;
1194
1195 desc.low = QI_CC_FM(fm) | QI_CC_SID(sid) | QI_CC_DID(did)
1196 | QI_CC_GRAN(type) | QI_CC_TYPE;
1197 desc.high = 0;
1198
1199 qi_submit_sync(&desc, iommu);
1200 }
1201
1202 void qi_flush_iotlb(struct intel_iommu *iommu, u16 did, u64 addr,
1203 unsigned int size_order, u64 type)
1204 {
1205 u8 dw = 0, dr = 0;
1206
1207 struct qi_desc desc;
1208 int ih = 0;
1209
1210 if (cap_write_drain(iommu->cap))
1211 dw = 1;
1212
1213 if (cap_read_drain(iommu->cap))
1214 dr = 1;
1215
1216 desc.low = QI_IOTLB_DID(did) | QI_IOTLB_DR(dr) | QI_IOTLB_DW(dw)
1217 | QI_IOTLB_GRAN(type) | QI_IOTLB_TYPE;
1218 desc.high = QI_IOTLB_ADDR(addr) | QI_IOTLB_IH(ih)
1219 | QI_IOTLB_AM(size_order);
1220
1221 qi_submit_sync(&desc, iommu);
1222 }
1223
1224 void qi_flush_dev_iotlb(struct intel_iommu *iommu, u16 sid, u16 qdep,
1225 u64 addr, unsigned mask)
1226 {
1227 struct qi_desc desc;
1228
1229 if (mask) {
1230 BUG_ON(addr & ((1 << (VTD_PAGE_SHIFT + mask)) - 1));
1231 addr |= (1 << (VTD_PAGE_SHIFT + mask - 1)) - 1;
1232 desc.high = QI_DEV_IOTLB_ADDR(addr) | QI_DEV_IOTLB_SIZE;
1233 } else
1234 desc.high = QI_DEV_IOTLB_ADDR(addr);
1235
1236 if (qdep >= QI_DEV_IOTLB_MAX_INVS)
1237 qdep = 0;
1238
1239 desc.low = QI_DEV_IOTLB_SID(sid) | QI_DEV_IOTLB_QDEP(qdep) |
1240 QI_DIOTLB_TYPE;
1241
1242 qi_submit_sync(&desc, iommu);
1243 }
1244
1245 /*
1246 * Disable Queued Invalidation interface.
1247 */
1248 void dmar_disable_qi(struct intel_iommu *iommu)
1249 {
1250 unsigned long flags;
1251 u32 sts;
1252 cycles_t start_time = get_cycles();
1253
1254 if (!ecap_qis(iommu->ecap))
1255 return;
1256
1257 raw_spin_lock_irqsave(&iommu->register_lock, flags);
1258
1259 sts = dmar_readq(iommu->reg + DMAR_GSTS_REG);
1260 if (!(sts & DMA_GSTS_QIES))
1261 goto end;
1262
1263 /*
1264 * Give a chance to HW to complete the pending invalidation requests.
1265 */
1266 while ((readl(iommu->reg + DMAR_IQT_REG) !=
1267 readl(iommu->reg + DMAR_IQH_REG)) &&
1268 (DMAR_OPERATION_TIMEOUT > (get_cycles() - start_time)))
1269 cpu_relax();
1270
1271 iommu->gcmd &= ~DMA_GCMD_QIE;
1272 writel(iommu->gcmd, iommu->reg + DMAR_GCMD_REG);
1273
1274 IOMMU_WAIT_OP(iommu, DMAR_GSTS_REG, readl,
1275 !(sts & DMA_GSTS_QIES), sts);
1276 end:
1277 raw_spin_unlock_irqrestore(&iommu->register_lock, flags);
1278 }
1279
1280 /*
1281 * Enable queued invalidation.
1282 */
1283 static void __dmar_enable_qi(struct intel_iommu *iommu)
1284 {
1285 u32 sts;
1286 unsigned long flags;
1287 struct q_inval *qi = iommu->qi;
1288
1289 qi->free_head = qi->free_tail = 0;
1290 qi->free_cnt = QI_LENGTH;
1291
1292 raw_spin_lock_irqsave(&iommu->register_lock, flags);
1293
1294 /* write zero to the tail reg */
1295 writel(0, iommu->reg + DMAR_IQT_REG);
1296
1297 dmar_writeq(iommu->reg + DMAR_IQA_REG, virt_to_phys(qi->desc));
1298
1299 iommu->gcmd |= DMA_GCMD_QIE;
1300 writel(iommu->gcmd, iommu->reg + DMAR_GCMD_REG);
1301
1302 /* Make sure hardware complete it */
1303 IOMMU_WAIT_OP(iommu, DMAR_GSTS_REG, readl, (sts & DMA_GSTS_QIES), sts);
1304
1305 raw_spin_unlock_irqrestore(&iommu->register_lock, flags);
1306 }
1307
1308 /*
1309 * Enable Queued Invalidation interface. This is a must to support
1310 * interrupt-remapping. Also used by DMA-remapping, which replaces
1311 * register based IOTLB invalidation.
1312 */
1313 int dmar_enable_qi(struct intel_iommu *iommu)
1314 {
1315 struct q_inval *qi;
1316 struct page *desc_page;
1317
1318 if (!ecap_qis(iommu->ecap))
1319 return -ENOENT;
1320
1321 /*
1322 * queued invalidation is already setup and enabled.
1323 */
1324 if (iommu->qi)
1325 return 0;
1326
1327 iommu->qi = kmalloc(sizeof(*qi), GFP_ATOMIC);
1328 if (!iommu->qi)
1329 return -ENOMEM;
1330
1331 qi = iommu->qi;
1332
1333
1334 desc_page = alloc_pages_node(iommu->node, GFP_ATOMIC | __GFP_ZERO, 0);
1335 if (!desc_page) {
1336 kfree(qi);
1337 iommu->qi = NULL;
1338 return -ENOMEM;
1339 }
1340
1341 qi->desc = page_address(desc_page);
1342
1343 qi->desc_status = kzalloc(QI_LENGTH * sizeof(int), GFP_ATOMIC);
1344 if (!qi->desc_status) {
1345 free_page((unsigned long) qi->desc);
1346 kfree(qi);
1347 iommu->qi = NULL;
1348 return -ENOMEM;
1349 }
1350
1351 qi->free_head = qi->free_tail = 0;
1352 qi->free_cnt = QI_LENGTH;
1353
1354 raw_spin_lock_init(&qi->q_lock);
1355
1356 __dmar_enable_qi(iommu);
1357
1358 return 0;
1359 }
1360
1361 /* iommu interrupt handling. Most stuff are MSI-like. */
1362
1363 enum faulttype {
1364 DMA_REMAP,
1365 INTR_REMAP,
1366 UNKNOWN,
1367 };
1368
1369 static const char *dma_remap_fault_reasons[] =
1370 {
1371 "Software",
1372 "Present bit in root entry is clear",
1373 "Present bit in context entry is clear",
1374 "Invalid context entry",
1375 "Access beyond MGAW",
1376 "PTE Write access is not set",
1377 "PTE Read access is not set",
1378 "Next page table ptr is invalid",
1379 "Root table address invalid",
1380 "Context table ptr is invalid",
1381 "non-zero reserved fields in RTP",
1382 "non-zero reserved fields in CTP",
1383 "non-zero reserved fields in PTE",
1384 "PCE for translation request specifies blocking",
1385 };
1386
1387 static const char *irq_remap_fault_reasons[] =
1388 {
1389 "Detected reserved fields in the decoded interrupt-remapped request",
1390 "Interrupt index exceeded the interrupt-remapping table size",
1391 "Present field in the IRTE entry is clear",
1392 "Error accessing interrupt-remapping table pointed by IRTA_REG",
1393 "Detected reserved fields in the IRTE entry",
1394 "Blocked a compatibility format interrupt request",
1395 "Blocked an interrupt request due to source-id verification failure",
1396 };
1397
1398 static const char *dmar_get_fault_reason(u8 fault_reason, int *fault_type)
1399 {
1400 if (fault_reason >= 0x20 && (fault_reason - 0x20 <
1401 ARRAY_SIZE(irq_remap_fault_reasons))) {
1402 *fault_type = INTR_REMAP;
1403 return irq_remap_fault_reasons[fault_reason - 0x20];
1404 } else if (fault_reason < ARRAY_SIZE(dma_remap_fault_reasons)) {
1405 *fault_type = DMA_REMAP;
1406 return dma_remap_fault_reasons[fault_reason];
1407 } else {
1408 *fault_type = UNKNOWN;
1409 return "Unknown";
1410 }
1411 }
1412
1413 void dmar_msi_unmask(struct irq_data *data)
1414 {
1415 struct intel_iommu *iommu = irq_data_get_irq_handler_data(data);
1416 unsigned long flag;
1417
1418 /* unmask it */
1419 raw_spin_lock_irqsave(&iommu->register_lock, flag);
1420 writel(0, iommu->reg + DMAR_FECTL_REG);
1421 /* Read a reg to force flush the post write */
1422 readl(iommu->reg + DMAR_FECTL_REG);
1423 raw_spin_unlock_irqrestore(&iommu->register_lock, flag);
1424 }
1425
1426 void dmar_msi_mask(struct irq_data *data)
1427 {
1428 unsigned long flag;
1429 struct intel_iommu *iommu = irq_data_get_irq_handler_data(data);
1430
1431 /* mask it */
1432 raw_spin_lock_irqsave(&iommu->register_lock, flag);
1433 writel(DMA_FECTL_IM, iommu->reg + DMAR_FECTL_REG);
1434 /* Read a reg to force flush the post write */
1435 readl(iommu->reg + DMAR_FECTL_REG);
1436 raw_spin_unlock_irqrestore(&iommu->register_lock, flag);
1437 }
1438
1439 void dmar_msi_write(int irq, struct msi_msg *msg)
1440 {
1441 struct intel_iommu *iommu = irq_get_handler_data(irq);
1442 unsigned long flag;
1443
1444 raw_spin_lock_irqsave(&iommu->register_lock, flag);
1445 writel(msg->data, iommu->reg + DMAR_FEDATA_REG);
1446 writel(msg->address_lo, iommu->reg + DMAR_FEADDR_REG);
1447 writel(msg->address_hi, iommu->reg + DMAR_FEUADDR_REG);
1448 raw_spin_unlock_irqrestore(&iommu->register_lock, flag);
1449 }
1450
1451 void dmar_msi_read(int irq, struct msi_msg *msg)
1452 {
1453 struct intel_iommu *iommu = irq_get_handler_data(irq);
1454 unsigned long flag;
1455
1456 raw_spin_lock_irqsave(&iommu->register_lock, flag);
1457 msg->data = readl(iommu->reg + DMAR_FEDATA_REG);
1458 msg->address_lo = readl(iommu->reg + DMAR_FEADDR_REG);
1459 msg->address_hi = readl(iommu->reg + DMAR_FEUADDR_REG);
1460 raw_spin_unlock_irqrestore(&iommu->register_lock, flag);
1461 }
1462
1463 static int dmar_fault_do_one(struct intel_iommu *iommu, int type,
1464 u8 fault_reason, u16 source_id, unsigned long long addr)
1465 {
1466 const char *reason;
1467 int fault_type;
1468
1469 reason = dmar_get_fault_reason(fault_reason, &fault_type);
1470
1471 if (fault_type == INTR_REMAP)
1472 pr_err("INTR-REMAP: Request device [[%02x:%02x.%d] "
1473 "fault index %llx\n"
1474 "INTR-REMAP:[fault reason %02d] %s\n",
1475 (source_id >> 8), PCI_SLOT(source_id & 0xFF),
1476 PCI_FUNC(source_id & 0xFF), addr >> 48,
1477 fault_reason, reason);
1478 else
1479 pr_err("DMAR:[%s] Request device [%02x:%02x.%d] "
1480 "fault addr %llx \n"
1481 "DMAR:[fault reason %02d] %s\n",
1482 (type ? "DMA Read" : "DMA Write"),
1483 (source_id >> 8), PCI_SLOT(source_id & 0xFF),
1484 PCI_FUNC(source_id & 0xFF), addr, fault_reason, reason);
1485 return 0;
1486 }
1487
1488 #define PRIMARY_FAULT_REG_LEN (16)
1489 irqreturn_t dmar_fault(int irq, void *dev_id)
1490 {
1491 struct intel_iommu *iommu = dev_id;
1492 int reg, fault_index;
1493 u32 fault_status;
1494 unsigned long flag;
1495
1496 raw_spin_lock_irqsave(&iommu->register_lock, flag);
1497 fault_status = readl(iommu->reg + DMAR_FSTS_REG);
1498 if (fault_status)
1499 pr_err("DRHD: handling fault status reg %x\n", fault_status);
1500
1501 /* TBD: ignore advanced fault log currently */
1502 if (!(fault_status & DMA_FSTS_PPF))
1503 goto unlock_exit;
1504
1505 fault_index = dma_fsts_fault_record_index(fault_status);
1506 reg = cap_fault_reg_offset(iommu->cap);
1507 while (1) {
1508 u8 fault_reason;
1509 u16 source_id;
1510 u64 guest_addr;
1511 int type;
1512 u32 data;
1513
1514 /* highest 32 bits */
1515 data = readl(iommu->reg + reg +
1516 fault_index * PRIMARY_FAULT_REG_LEN + 12);
1517 if (!(data & DMA_FRCD_F))
1518 break;
1519
1520 fault_reason = dma_frcd_fault_reason(data);
1521 type = dma_frcd_type(data);
1522
1523 data = readl(iommu->reg + reg +
1524 fault_index * PRIMARY_FAULT_REG_LEN + 8);
1525 source_id = dma_frcd_source_id(data);
1526
1527 guest_addr = dmar_readq(iommu->reg + reg +
1528 fault_index * PRIMARY_FAULT_REG_LEN);
1529 guest_addr = dma_frcd_page_addr(guest_addr);
1530 /* clear the fault */
1531 writel(DMA_FRCD_F, iommu->reg + reg +
1532 fault_index * PRIMARY_FAULT_REG_LEN + 12);
1533
1534 raw_spin_unlock_irqrestore(&iommu->register_lock, flag);
1535
1536 dmar_fault_do_one(iommu, type, fault_reason,
1537 source_id, guest_addr);
1538
1539 fault_index++;
1540 if (fault_index >= cap_num_fault_regs(iommu->cap))
1541 fault_index = 0;
1542 raw_spin_lock_irqsave(&iommu->register_lock, flag);
1543 }
1544
1545 writel(DMA_FSTS_PFO | DMA_FSTS_PPF, iommu->reg + DMAR_FSTS_REG);
1546
1547 unlock_exit:
1548 raw_spin_unlock_irqrestore(&iommu->register_lock, flag);
1549 return IRQ_HANDLED;
1550 }
1551
1552 int dmar_set_interrupt(struct intel_iommu *iommu)
1553 {
1554 int irq, ret;
1555
1556 /*
1557 * Check if the fault interrupt is already initialized.
1558 */
1559 if (iommu->irq)
1560 return 0;
1561
1562 irq = dmar_alloc_hwirq();
1563 if (irq <= 0) {
1564 pr_err("IOMMU: no free vectors\n");
1565 return -EINVAL;
1566 }
1567
1568 irq_set_handler_data(irq, iommu);
1569 iommu->irq = irq;
1570
1571 ret = arch_setup_dmar_msi(irq);
1572 if (ret) {
1573 irq_set_handler_data(irq, NULL);
1574 iommu->irq = 0;
1575 dmar_free_hwirq(irq);
1576 return ret;
1577 }
1578
1579 ret = request_irq(irq, dmar_fault, IRQF_NO_THREAD, iommu->name, iommu);
1580 if (ret)
1581 pr_err("IOMMU: can't request irq\n");
1582 return ret;
1583 }
1584
1585 int __init enable_drhd_fault_handling(void)
1586 {
1587 struct dmar_drhd_unit *drhd;
1588 struct intel_iommu *iommu;
1589
1590 /*
1591 * Enable fault control interrupt.
1592 */
1593 for_each_iommu(iommu, drhd) {
1594 u32 fault_status;
1595 int ret = dmar_set_interrupt(iommu);
1596
1597 if (ret) {
1598 pr_err("DRHD %Lx: failed to enable fault, interrupt, ret %d\n",
1599 (unsigned long long)drhd->reg_base_addr, ret);
1600 return -1;
1601 }
1602
1603 /*
1604 * Clear any previous faults.
1605 */
1606 dmar_fault(iommu->irq, iommu);
1607 fault_status = readl(iommu->reg + DMAR_FSTS_REG);
1608 writel(fault_status, iommu->reg + DMAR_FSTS_REG);
1609 }
1610
1611 return 0;
1612 }
1613
1614 /*
1615 * Re-enable Queued Invalidation interface.
1616 */
1617 int dmar_reenable_qi(struct intel_iommu *iommu)
1618 {
1619 if (!ecap_qis(iommu->ecap))
1620 return -ENOENT;
1621
1622 if (!iommu->qi)
1623 return -ENOENT;
1624
1625 /*
1626 * First disable queued invalidation.
1627 */
1628 dmar_disable_qi(iommu);
1629 /*
1630 * Then enable queued invalidation again. Since there is no pending
1631 * invalidation requests now, it's safe to re-enable queued
1632 * invalidation.
1633 */
1634 __dmar_enable_qi(iommu);
1635
1636 return 0;
1637 }
1638
1639 /*
1640 * Check interrupt remapping support in DMAR table description.
1641 */
1642 int __init dmar_ir_support(void)
1643 {
1644 struct acpi_table_dmar *dmar;
1645 dmar = (struct acpi_table_dmar *)dmar_tbl;
1646 if (!dmar)
1647 return 0;
1648 return dmar->flags & 0x1;
1649 }
1650
1651 static int __init dmar_free_unused_resources(void)
1652 {
1653 struct dmar_drhd_unit *dmaru, *dmaru_n;
1654
1655 /* DMAR units are in use */
1656 if (irq_remapping_enabled || intel_iommu_enabled)
1657 return 0;
1658
1659 if (dmar_dev_scope_status != 1 && !list_empty(&dmar_drhd_units))
1660 bus_unregister_notifier(&pci_bus_type, &dmar_pci_bus_nb);
1661
1662 down_write(&dmar_global_lock);
1663 list_for_each_entry_safe(dmaru, dmaru_n, &dmar_drhd_units, list) {
1664 list_del(&dmaru->list);
1665 dmar_free_drhd(dmaru);
1666 }
1667 up_write(&dmar_global_lock);
1668
1669 return 0;
1670 }
1671
1672 late_initcall(dmar_free_unused_resources);
1673 IOMMU_INIT_POST(detect_intel_iommu);
Linux kernel - Deliverables
This page took 0.067631 seconds and 6 git commands to generate.