Loading...
Note: File does not exist in v5.4.
1// SPDX-License-Identifier: GPL-2.0-only
2/*
3 * Copyright (c) 2006, Intel Corporation.
4 *
5 * Copyright (C) 2006-2008 Intel Corporation
6 * Author: Ashok Raj <ashok.raj@intel.com>
7 * Author: Shaohua Li <shaohua.li@intel.com>
8 * Author: Anil S Keshavamurthy <anil.s.keshavamurthy@intel.com>
9 *
10 * This file implements early detection/parsing of Remapping Devices
11 * reported to OS through BIOS via DMA remapping reporting (DMAR) ACPI
12 * tables.
13 *
14 * These routines are used by both DMA-remapping and Interrupt-remapping
15 */
16
17#define pr_fmt(fmt) "DMAR: " fmt
18
19#include <linux/pci.h>
20#include <linux/dmar.h>
21#include <linux/iova.h>
22#include <linux/timer.h>
23#include <linux/irq.h>
24#include <linux/interrupt.h>
25#include <linux/tboot.h>
26#include <linux/dmi.h>
27#include <linux/slab.h>
28#include <linux/iommu.h>
29#include <linux/numa.h>
30#include <linux/limits.h>
31#include <asm/irq_remapping.h>
32
33#include "iommu.h"
34#include "../irq_remapping.h"
35#include "perf.h"
36#include "trace.h"
37#include "perfmon.h"
38
39typedef int (*dmar_res_handler_t)(struct acpi_dmar_header *, void *);
40struct dmar_res_callback {
41 dmar_res_handler_t cb[ACPI_DMAR_TYPE_RESERVED];
42 void *arg[ACPI_DMAR_TYPE_RESERVED];
43 bool ignore_unhandled;
44 bool print_entry;
45};
46
47/*
48 * Assumptions:
49 * 1) The hotplug framework guarentees that DMAR unit will be hot-added
50 * before IO devices managed by that unit.
51 * 2) The hotplug framework guarantees that DMAR unit will be hot-removed
52 * after IO devices managed by that unit.
53 * 3) Hotplug events are rare.
54 *
55 * Locking rules for DMA and interrupt remapping related global data structures:
56 * 1) Use dmar_global_lock in process context
57 * 2) Use RCU in interrupt context
58 */
59DECLARE_RWSEM(dmar_global_lock);
60LIST_HEAD(dmar_drhd_units);
61
62struct acpi_table_header * __initdata dmar_tbl;
63static int dmar_dev_scope_status = 1;
64static DEFINE_IDA(dmar_seq_ids);
65
66static int alloc_iommu(struct dmar_drhd_unit *drhd);
67static void free_iommu(struct intel_iommu *iommu);
68
69static void 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
81void *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_NAMESPACE ||
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
104void 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 */
120static char dmar_pci_notify_info_buf[64];
121
122static struct dmar_pci_notify_info *
123dmar_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 /*
131 * Ignore devices that have a domain number higher than what can
132 * be looked up in DMAR, e.g. VMD subdevices with domain 0x10000
133 */
134 if (pci_domain_nr(dev->bus) > U16_MAX)
135 return NULL;
136
137 /* Only generate path[] for device addition event */
138 if (event == BUS_NOTIFY_ADD_DEVICE)
139 for (tmp = dev; tmp; tmp = tmp->bus->self)
140 level++;
141
142 size = struct_size(info, path, level);
143 if (size <= sizeof(dmar_pci_notify_info_buf)) {
144 info = (struct dmar_pci_notify_info *)dmar_pci_notify_info_buf;
145 } else {
146 info = kzalloc(size, GFP_KERNEL);
147 if (!info) {
148 if (dmar_dev_scope_status == 0)
149 dmar_dev_scope_status = -ENOMEM;
150 return NULL;
151 }
152 }
153
154 info->event = event;
155 info->dev = dev;
156 info->seg = pci_domain_nr(dev->bus);
157 info->level = level;
158 if (event == BUS_NOTIFY_ADD_DEVICE) {
159 for (tmp = dev; tmp; tmp = tmp->bus->self) {
160 level--;
161 info->path[level].bus = tmp->bus->number;
162 info->path[level].device = PCI_SLOT(tmp->devfn);
163 info->path[level].function = PCI_FUNC(tmp->devfn);
164 if (pci_is_root_bus(tmp->bus))
165 info->bus = tmp->bus->number;
166 }
167 }
168
169 return info;
170}
171
172static inline void dmar_free_pci_notify_info(struct dmar_pci_notify_info *info)
173{
174 if ((void *)info != dmar_pci_notify_info_buf)
175 kfree(info);
176}
177
178static bool dmar_match_pci_path(struct dmar_pci_notify_info *info, int bus,
179 struct acpi_dmar_pci_path *path, int count)
180{
181 int i;
182
183 if (info->bus != bus)
184 goto fallback;
185 if (info->level != count)
186 goto fallback;
187
188 for (i = 0; i < count; i++) {
189 if (path[i].device != info->path[i].device ||
190 path[i].function != info->path[i].function)
191 goto fallback;
192 }
193
194 return true;
195
196fallback:
197
198 if (count != 1)
199 return false;
200
201 i = info->level - 1;
202 if (bus == info->path[i].bus &&
203 path[0].device == info->path[i].device &&
204 path[0].function == info->path[i].function) {
205 pr_info(FW_BUG "RMRR entry for device %02x:%02x.%x is broken - applying workaround\n",
206 bus, path[0].device, path[0].function);
207 return true;
208 }
209
210 return false;
211}
212
213/* Return: > 0 if match found, 0 if no match found, < 0 if error happens */
214int dmar_insert_dev_scope(struct dmar_pci_notify_info *info,
215 void *start, void*end, u16 segment,
216 struct dmar_dev_scope *devices,
217 int devices_cnt)
218{
219 int i, level;
220 struct device *tmp, *dev = &info->dev->dev;
221 struct acpi_dmar_device_scope *scope;
222 struct acpi_dmar_pci_path *path;
223
224 if (segment != info->seg)
225 return 0;
226
227 for (; start < end; start += scope->length) {
228 scope = start;
229 if (scope->entry_type != ACPI_DMAR_SCOPE_TYPE_ENDPOINT &&
230 scope->entry_type != ACPI_DMAR_SCOPE_TYPE_BRIDGE)
231 continue;
232
233 path = (struct acpi_dmar_pci_path *)(scope + 1);
234 level = (scope->length - sizeof(*scope)) / sizeof(*path);
235 if (!dmar_match_pci_path(info, scope->bus, path, level))
236 continue;
237
238 /*
239 * We expect devices with endpoint scope to have normal PCI
240 * headers, and devices with bridge scope to have bridge PCI
241 * headers. However PCI NTB devices may be listed in the
242 * DMAR table with bridge scope, even though they have a
243 * normal PCI header. NTB devices are identified by class
244 * "BRIDGE_OTHER" (0680h) - we don't declare a socpe mismatch
245 * for this special case.
246 */
247 if ((scope->entry_type == ACPI_DMAR_SCOPE_TYPE_ENDPOINT &&
248 info->dev->hdr_type != PCI_HEADER_TYPE_NORMAL) ||
249 (scope->entry_type == ACPI_DMAR_SCOPE_TYPE_BRIDGE &&
250 (info->dev->hdr_type == PCI_HEADER_TYPE_NORMAL &&
251 info->dev->class >> 16 != PCI_BASE_CLASS_BRIDGE))) {
252 pr_warn("Device scope type does not match for %s\n",
253 pci_name(info->dev));
254 return -EINVAL;
255 }
256
257 for_each_dev_scope(devices, devices_cnt, i, tmp)
258 if (tmp == NULL) {
259 devices[i].bus = info->dev->bus->number;
260 devices[i].devfn = info->dev->devfn;
261 rcu_assign_pointer(devices[i].dev,
262 get_device(dev));
263 return 1;
264 }
265 if (WARN_ON(i >= devices_cnt))
266 return -EINVAL;
267 }
268
269 return 0;
270}
271
272int dmar_remove_dev_scope(struct dmar_pci_notify_info *info, u16 segment,
273 struct dmar_dev_scope *devices, int count)
274{
275 int index;
276 struct device *tmp;
277
278 if (info->seg != segment)
279 return 0;
280
281 for_each_active_dev_scope(devices, count, index, tmp)
282 if (tmp == &info->dev->dev) {
283 RCU_INIT_POINTER(devices[index].dev, NULL);
284 synchronize_rcu();
285 put_device(tmp);
286 return 1;
287 }
288
289 return 0;
290}
291
292static int dmar_pci_bus_add_dev(struct dmar_pci_notify_info *info)
293{
294 int ret = 0;
295 struct dmar_drhd_unit *dmaru;
296 struct acpi_dmar_hardware_unit *drhd;
297
298 for_each_drhd_unit(dmaru) {
299 if (dmaru->include_all)
300 continue;
301
302 drhd = container_of(dmaru->hdr,
303 struct acpi_dmar_hardware_unit, header);
304 ret = dmar_insert_dev_scope(info, (void *)(drhd + 1),
305 ((void *)drhd) + drhd->header.length,
306 dmaru->segment,
307 dmaru->devices, dmaru->devices_cnt);
308 if (ret)
309 break;
310 }
311 if (ret >= 0)
312 ret = dmar_iommu_notify_scope_dev(info);
313 if (ret < 0 && dmar_dev_scope_status == 0)
314 dmar_dev_scope_status = ret;
315
316 if (ret >= 0)
317 intel_irq_remap_add_device(info);
318
319 return ret;
320}
321
322static void dmar_pci_bus_del_dev(struct dmar_pci_notify_info *info)
323{
324 struct dmar_drhd_unit *dmaru;
325
326 for_each_drhd_unit(dmaru)
327 if (dmar_remove_dev_scope(info, dmaru->segment,
328 dmaru->devices, dmaru->devices_cnt))
329 break;
330 dmar_iommu_notify_scope_dev(info);
331}
332
333static inline void vf_inherit_msi_domain(struct pci_dev *pdev)
334{
335 struct pci_dev *physfn = pci_physfn(pdev);
336
337 dev_set_msi_domain(&pdev->dev, dev_get_msi_domain(&physfn->dev));
338}
339
340static int dmar_pci_bus_notifier(struct notifier_block *nb,
341 unsigned long action, void *data)
342{
343 struct pci_dev *pdev = to_pci_dev(data);
344 struct dmar_pci_notify_info *info;
345
346 /* Only care about add/remove events for physical functions.
347 * For VFs we actually do the lookup based on the corresponding
348 * PF in device_to_iommu() anyway. */
349 if (pdev->is_virtfn) {
350 /*
351 * Ensure that the VF device inherits the irq domain of the
352 * PF device. Ideally the device would inherit the domain
353 * from the bus, but DMAR can have multiple units per bus
354 * which makes this impossible. The VF 'bus' could inherit
355 * from the PF device, but that's yet another x86'sism to
356 * inflict on everybody else.
357 */
358 if (action == BUS_NOTIFY_ADD_DEVICE)
359 vf_inherit_msi_domain(pdev);
360 return NOTIFY_DONE;
361 }
362
363 if (action != BUS_NOTIFY_ADD_DEVICE &&
364 action != BUS_NOTIFY_REMOVED_DEVICE)
365 return NOTIFY_DONE;
366
367 info = dmar_alloc_pci_notify_info(pdev, action);
368 if (!info)
369 return NOTIFY_DONE;
370
371 down_write(&dmar_global_lock);
372 if (action == BUS_NOTIFY_ADD_DEVICE)
373 dmar_pci_bus_add_dev(info);
374 else if (action == BUS_NOTIFY_REMOVED_DEVICE)
375 dmar_pci_bus_del_dev(info);
376 up_write(&dmar_global_lock);
377
378 dmar_free_pci_notify_info(info);
379
380 return NOTIFY_OK;
381}
382
383static struct notifier_block dmar_pci_bus_nb = {
384 .notifier_call = dmar_pci_bus_notifier,
385 .priority = 1,
386};
387
388static struct dmar_drhd_unit *
389dmar_find_dmaru(struct acpi_dmar_hardware_unit *drhd)
390{
391 struct dmar_drhd_unit *dmaru;
392
393 list_for_each_entry_rcu(dmaru, &dmar_drhd_units, list,
394 dmar_rcu_check())
395 if (dmaru->segment == drhd->segment &&
396 dmaru->reg_base_addr == drhd->address)
397 return dmaru;
398
399 return NULL;
400}
401
402/*
403 * dmar_parse_one_drhd - parses exactly one DMA remapping hardware definition
404 * structure which uniquely represent one DMA remapping hardware unit
405 * present in the platform
406 */
407static int dmar_parse_one_drhd(struct acpi_dmar_header *header, void *arg)
408{
409 struct acpi_dmar_hardware_unit *drhd;
410 struct dmar_drhd_unit *dmaru;
411 int ret;
412
413 drhd = (struct acpi_dmar_hardware_unit *)header;
414 dmaru = dmar_find_dmaru(drhd);
415 if (dmaru)
416 goto out;
417
418 dmaru = kzalloc(sizeof(*dmaru) + header->length, GFP_KERNEL);
419 if (!dmaru)
420 return -ENOMEM;
421
422 /*
423 * If header is allocated from slab by ACPI _DSM method, we need to
424 * copy the content because the memory buffer will be freed on return.
425 */
426 dmaru->hdr = (void *)(dmaru + 1);
427 memcpy(dmaru->hdr, header, header->length);
428 dmaru->reg_base_addr = drhd->address;
429 dmaru->segment = drhd->segment;
430 /* The size of the register set is 2 ^ N 4 KB pages. */
431 dmaru->reg_size = 1UL << (drhd->size + 12);
432 dmaru->include_all = drhd->flags & 0x1; /* BIT0: INCLUDE_ALL */
433 dmaru->devices = dmar_alloc_dev_scope((void *)(drhd + 1),
434 ((void *)drhd) + drhd->header.length,
435 &dmaru->devices_cnt);
436 if (dmaru->devices_cnt && dmaru->devices == NULL) {
437 kfree(dmaru);
438 return -ENOMEM;
439 }
440
441 ret = alloc_iommu(dmaru);
442 if (ret) {
443 dmar_free_dev_scope(&dmaru->devices,
444 &dmaru->devices_cnt);
445 kfree(dmaru);
446 return ret;
447 }
448 dmar_register_drhd_unit(dmaru);
449
450out:
451 if (arg)
452 (*(int *)arg)++;
453
454 return 0;
455}
456
457static void dmar_free_drhd(struct dmar_drhd_unit *dmaru)
458{
459 if (dmaru->devices && dmaru->devices_cnt)
460 dmar_free_dev_scope(&dmaru->devices, &dmaru->devices_cnt);
461 if (dmaru->iommu)
462 free_iommu(dmaru->iommu);
463 kfree(dmaru);
464}
465
466static int __init dmar_parse_one_andd(struct acpi_dmar_header *header,
467 void *arg)
468{
469 struct acpi_dmar_andd *andd = (void *)header;
470
471 /* Check for NUL termination within the designated length */
472 if (strnlen(andd->device_name, header->length - 8) == header->length - 8) {
473 pr_warn(FW_BUG
474 "Your BIOS is broken; ANDD object name is not NUL-terminated\n"
475 "BIOS vendor: %s; Ver: %s; Product Version: %s\n",
476 dmi_get_system_info(DMI_BIOS_VENDOR),
477 dmi_get_system_info(DMI_BIOS_VERSION),
478 dmi_get_system_info(DMI_PRODUCT_VERSION));
479 add_taint(TAINT_FIRMWARE_WORKAROUND, LOCKDEP_STILL_OK);
480 return -EINVAL;
481 }
482 pr_info("ANDD device: %x name: %s\n", andd->device_number,
483 andd->device_name);
484
485 return 0;
486}
487
488#ifdef CONFIG_ACPI_NUMA
489static int dmar_parse_one_rhsa(struct acpi_dmar_header *header, void *arg)
490{
491 struct acpi_dmar_rhsa *rhsa;
492 struct dmar_drhd_unit *drhd;
493
494 rhsa = (struct acpi_dmar_rhsa *)header;
495 for_each_drhd_unit(drhd) {
496 if (drhd->reg_base_addr == rhsa->base_address) {
497 int node = pxm_to_node(rhsa->proximity_domain);
498
499 if (node != NUMA_NO_NODE && !node_online(node))
500 node = NUMA_NO_NODE;
501 drhd->iommu->node = node;
502 return 0;
503 }
504 }
505 pr_warn(FW_BUG
506 "Your BIOS is broken; RHSA refers to non-existent DMAR unit at %llx\n"
507 "BIOS vendor: %s; Ver: %s; Product Version: %s\n",
508 rhsa->base_address,
509 dmi_get_system_info(DMI_BIOS_VENDOR),
510 dmi_get_system_info(DMI_BIOS_VERSION),
511 dmi_get_system_info(DMI_PRODUCT_VERSION));
512 add_taint(TAINT_FIRMWARE_WORKAROUND, LOCKDEP_STILL_OK);
513
514 return 0;
515}
516#else
517#define dmar_parse_one_rhsa dmar_res_noop
518#endif
519
520static void
521dmar_table_print_dmar_entry(struct acpi_dmar_header *header)
522{
523 struct acpi_dmar_hardware_unit *drhd;
524 struct acpi_dmar_reserved_memory *rmrr;
525 struct acpi_dmar_atsr *atsr;
526 struct acpi_dmar_rhsa *rhsa;
527 struct acpi_dmar_satc *satc;
528
529 switch (header->type) {
530 case ACPI_DMAR_TYPE_HARDWARE_UNIT:
531 drhd = container_of(header, struct acpi_dmar_hardware_unit,
532 header);
533 pr_info("DRHD base: %#016Lx flags: %#x\n",
534 (unsigned long long)drhd->address, drhd->flags);
535 break;
536 case ACPI_DMAR_TYPE_RESERVED_MEMORY:
537 rmrr = container_of(header, struct acpi_dmar_reserved_memory,
538 header);
539 pr_info("RMRR base: %#016Lx end: %#016Lx\n",
540 (unsigned long long)rmrr->base_address,
541 (unsigned long long)rmrr->end_address);
542 break;
543 case ACPI_DMAR_TYPE_ROOT_ATS:
544 atsr = container_of(header, struct acpi_dmar_atsr, header);
545 pr_info("ATSR flags: %#x\n", atsr->flags);
546 break;
547 case ACPI_DMAR_TYPE_HARDWARE_AFFINITY:
548 rhsa = container_of(header, struct acpi_dmar_rhsa, header);
549 pr_info("RHSA base: %#016Lx proximity domain: %#x\n",
550 (unsigned long long)rhsa->base_address,
551 rhsa->proximity_domain);
552 break;
553 case ACPI_DMAR_TYPE_NAMESPACE:
554 /* We don't print this here because we need to sanity-check
555 it first. So print it in dmar_parse_one_andd() instead. */
556 break;
557 case ACPI_DMAR_TYPE_SATC:
558 satc = container_of(header, struct acpi_dmar_satc, header);
559 pr_info("SATC flags: 0x%x\n", satc->flags);
560 break;
561 }
562}
563
564/**
565 * dmar_table_detect - checks to see if the platform supports DMAR devices
566 */
567static int __init dmar_table_detect(void)
568{
569 acpi_status status = AE_OK;
570
571 /* if we could find DMAR table, then there are DMAR devices */
572 status = acpi_get_table(ACPI_SIG_DMAR, 0, &dmar_tbl);
573
574 if (ACPI_SUCCESS(status) && !dmar_tbl) {
575 pr_warn("Unable to map DMAR\n");
576 status = AE_NOT_FOUND;
577 }
578
579 return ACPI_SUCCESS(status) ? 0 : -ENOENT;
580}
581
582static int dmar_walk_remapping_entries(struct acpi_dmar_header *start,
583 size_t len, struct dmar_res_callback *cb)
584{
585 struct acpi_dmar_header *iter, *next;
586 struct acpi_dmar_header *end = ((void *)start) + len;
587
588 for (iter = start; iter < end; iter = next) {
589 next = (void *)iter + iter->length;
590 if (iter->length == 0) {
591 /* Avoid looping forever on bad ACPI tables */
592 pr_debug(FW_BUG "Invalid 0-length structure\n");
593 break;
594 } else if (next > end) {
595 /* Avoid passing table end */
596 pr_warn(FW_BUG "Record passes table end\n");
597 return -EINVAL;
598 }
599
600 if (cb->print_entry)
601 dmar_table_print_dmar_entry(iter);
602
603 if (iter->type >= ACPI_DMAR_TYPE_RESERVED) {
604 /* continue for forward compatibility */
605 pr_debug("Unknown DMAR structure type %d\n",
606 iter->type);
607 } else if (cb->cb[iter->type]) {
608 int ret;
609
610 ret = cb->cb[iter->type](iter, cb->arg[iter->type]);
611 if (ret)
612 return ret;
613 } else if (!cb->ignore_unhandled) {
614 pr_warn("No handler for DMAR structure type %d\n",
615 iter->type);
616 return -EINVAL;
617 }
618 }
619
620 return 0;
621}
622
623static inline int dmar_walk_dmar_table(struct acpi_table_dmar *dmar,
624 struct dmar_res_callback *cb)
625{
626 return dmar_walk_remapping_entries((void *)(dmar + 1),
627 dmar->header.length - sizeof(*dmar), cb);
628}
629
630/**
631 * parse_dmar_table - parses the DMA reporting table
632 */
633static int __init
634parse_dmar_table(void)
635{
636 struct acpi_table_dmar *dmar;
637 int drhd_count = 0;
638 int ret;
639 struct dmar_res_callback cb = {
640 .print_entry = true,
641 .ignore_unhandled = true,
642 .arg[ACPI_DMAR_TYPE_HARDWARE_UNIT] = &drhd_count,
643 .cb[ACPI_DMAR_TYPE_HARDWARE_UNIT] = &dmar_parse_one_drhd,
644 .cb[ACPI_DMAR_TYPE_RESERVED_MEMORY] = &dmar_parse_one_rmrr,
645 .cb[ACPI_DMAR_TYPE_ROOT_ATS] = &dmar_parse_one_atsr,
646 .cb[ACPI_DMAR_TYPE_HARDWARE_AFFINITY] = &dmar_parse_one_rhsa,
647 .cb[ACPI_DMAR_TYPE_NAMESPACE] = &dmar_parse_one_andd,
648 .cb[ACPI_DMAR_TYPE_SATC] = &dmar_parse_one_satc,
649 };
650
651 /*
652 * Do it again, earlier dmar_tbl mapping could be mapped with
653 * fixed map.
654 */
655 dmar_table_detect();
656
657 /*
658 * ACPI tables may not be DMA protected by tboot, so use DMAR copy
659 * SINIT saved in SinitMleData in TXT heap (which is DMA protected)
660 */
661 dmar_tbl = tboot_get_dmar_table(dmar_tbl);
662
663 dmar = (struct acpi_table_dmar *)dmar_tbl;
664 if (!dmar)
665 return -ENODEV;
666
667 if (dmar->width < PAGE_SHIFT - 1) {
668 pr_warn("Invalid DMAR haw\n");
669 return -EINVAL;
670 }
671
672 pr_info("Host address width %d\n", dmar->width + 1);
673 ret = dmar_walk_dmar_table(dmar, &cb);
674 if (ret == 0 && drhd_count == 0)
675 pr_warn(FW_BUG "No DRHD structure found in DMAR table\n");
676
677 return ret;
678}
679
680static int dmar_pci_device_match(struct dmar_dev_scope devices[],
681 int cnt, struct pci_dev *dev)
682{
683 int index;
684 struct device *tmp;
685
686 while (dev) {
687 for_each_active_dev_scope(devices, cnt, index, tmp)
688 if (dev_is_pci(tmp) && dev == to_pci_dev(tmp))
689 return 1;
690
691 /* Check our parent */
692 dev = dev->bus->self;
693 }
694
695 return 0;
696}
697
698struct dmar_drhd_unit *
699dmar_find_matched_drhd_unit(struct pci_dev *dev)
700{
701 struct dmar_drhd_unit *dmaru;
702 struct acpi_dmar_hardware_unit *drhd;
703
704 dev = pci_physfn(dev);
705
706 rcu_read_lock();
707 for_each_drhd_unit(dmaru) {
708 drhd = container_of(dmaru->hdr,
709 struct acpi_dmar_hardware_unit,
710 header);
711
712 if (dmaru->include_all &&
713 drhd->segment == pci_domain_nr(dev->bus))
714 goto out;
715
716 if (dmar_pci_device_match(dmaru->devices,
717 dmaru->devices_cnt, dev))
718 goto out;
719 }
720 dmaru = NULL;
721out:
722 rcu_read_unlock();
723
724 return dmaru;
725}
726
727static void __init dmar_acpi_insert_dev_scope(u8 device_number,
728 struct acpi_device *adev)
729{
730 struct dmar_drhd_unit *dmaru;
731 struct acpi_dmar_hardware_unit *drhd;
732 struct acpi_dmar_device_scope *scope;
733 struct device *tmp;
734 int i;
735 struct acpi_dmar_pci_path *path;
736
737 for_each_drhd_unit(dmaru) {
738 drhd = container_of(dmaru->hdr,
739 struct acpi_dmar_hardware_unit,
740 header);
741
742 for (scope = (void *)(drhd + 1);
743 (unsigned long)scope < ((unsigned long)drhd) + drhd->header.length;
744 scope = ((void *)scope) + scope->length) {
745 if (scope->entry_type != ACPI_DMAR_SCOPE_TYPE_NAMESPACE)
746 continue;
747 if (scope->enumeration_id != device_number)
748 continue;
749
750 path = (void *)(scope + 1);
751 pr_info("ACPI device \"%s\" under DMAR at %llx as %02x:%02x.%d\n",
752 dev_name(&adev->dev), dmaru->reg_base_addr,
753 scope->bus, path->device, path->function);
754 for_each_dev_scope(dmaru->devices, dmaru->devices_cnt, i, tmp)
755 if (tmp == NULL) {
756 dmaru->devices[i].bus = scope->bus;
757 dmaru->devices[i].devfn = PCI_DEVFN(path->device,
758 path->function);
759 rcu_assign_pointer(dmaru->devices[i].dev,
760 get_device(&adev->dev));
761 return;
762 }
763 BUG_ON(i >= dmaru->devices_cnt);
764 }
765 }
766 pr_warn("No IOMMU scope found for ANDD enumeration ID %d (%s)\n",
767 device_number, dev_name(&adev->dev));
768}
769
770static int __init dmar_acpi_dev_scope_init(void)
771{
772 struct acpi_dmar_andd *andd;
773
774 if (dmar_tbl == NULL)
775 return -ENODEV;
776
777 for (andd = (void *)dmar_tbl + sizeof(struct acpi_table_dmar);
778 ((unsigned long)andd) < ((unsigned long)dmar_tbl) + dmar_tbl->length;
779 andd = ((void *)andd) + andd->header.length) {
780 if (andd->header.type == ACPI_DMAR_TYPE_NAMESPACE) {
781 acpi_handle h;
782 struct acpi_device *adev;
783
784 if (!ACPI_SUCCESS(acpi_get_handle(ACPI_ROOT_OBJECT,
785 andd->device_name,
786 &h))) {
787 pr_err("Failed to find handle for ACPI object %s\n",
788 andd->device_name);
789 continue;
790 }
791 adev = acpi_fetch_acpi_dev(h);
792 if (!adev) {
793 pr_err("Failed to get device for ACPI object %s\n",
794 andd->device_name);
795 continue;
796 }
797 dmar_acpi_insert_dev_scope(andd->device_number, adev);
798 }
799 }
800 return 0;
801}
802
803int __init dmar_dev_scope_init(void)
804{
805 struct pci_dev *dev = NULL;
806 struct dmar_pci_notify_info *info;
807
808 if (dmar_dev_scope_status != 1)
809 return dmar_dev_scope_status;
810
811 if (list_empty(&dmar_drhd_units)) {
812 dmar_dev_scope_status = -ENODEV;
813 } else {
814 dmar_dev_scope_status = 0;
815
816 dmar_acpi_dev_scope_init();
817
818 for_each_pci_dev(dev) {
819 if (dev->is_virtfn)
820 continue;
821
822 info = dmar_alloc_pci_notify_info(dev,
823 BUS_NOTIFY_ADD_DEVICE);
824 if (!info) {
825 pci_dev_put(dev);
826 return dmar_dev_scope_status;
827 } else {
828 dmar_pci_bus_add_dev(info);
829 dmar_free_pci_notify_info(info);
830 }
831 }
832 }
833
834 return dmar_dev_scope_status;
835}
836
837void __init dmar_register_bus_notifier(void)
838{
839 bus_register_notifier(&pci_bus_type, &dmar_pci_bus_nb);
840}
841
842
843int __init dmar_table_init(void)
844{
845 static int dmar_table_initialized;
846 int ret;
847
848 if (dmar_table_initialized == 0) {
849 ret = parse_dmar_table();
850 if (ret < 0) {
851 if (ret != -ENODEV)
852 pr_info("Parse DMAR table failure.\n");
853 } else if (list_empty(&dmar_drhd_units)) {
854 pr_info("No DMAR devices found\n");
855 ret = -ENODEV;
856 }
857
858 if (ret < 0)
859 dmar_table_initialized = ret;
860 else
861 dmar_table_initialized = 1;
862 }
863
864 return dmar_table_initialized < 0 ? dmar_table_initialized : 0;
865}
866
867static void warn_invalid_dmar(u64 addr, const char *message)
868{
869 pr_warn_once(FW_BUG
870 "Your BIOS is broken; DMAR reported at address %llx%s!\n"
871 "BIOS vendor: %s; Ver: %s; Product Version: %s\n",
872 addr, message,
873 dmi_get_system_info(DMI_BIOS_VENDOR),
874 dmi_get_system_info(DMI_BIOS_VERSION),
875 dmi_get_system_info(DMI_PRODUCT_VERSION));
876 add_taint(TAINT_FIRMWARE_WORKAROUND, LOCKDEP_STILL_OK);
877}
878
879static int __ref
880dmar_validate_one_drhd(struct acpi_dmar_header *entry, void *arg)
881{
882 struct acpi_dmar_hardware_unit *drhd;
883 void __iomem *addr;
884 u64 cap, ecap;
885
886 drhd = (void *)entry;
887 if (!drhd->address) {
888 warn_invalid_dmar(0, "");
889 return -EINVAL;
890 }
891
892 if (arg)
893 addr = ioremap(drhd->address, VTD_PAGE_SIZE);
894 else
895 addr = early_ioremap(drhd->address, VTD_PAGE_SIZE);
896 if (!addr) {
897 pr_warn("Can't validate DRHD address: %llx\n", drhd->address);
898 return -EINVAL;
899 }
900
901 cap = dmar_readq(addr + DMAR_CAP_REG);
902 ecap = dmar_readq(addr + DMAR_ECAP_REG);
903
904 if (arg)
905 iounmap(addr);
906 else
907 early_iounmap(addr, VTD_PAGE_SIZE);
908
909 if (cap == (uint64_t)-1 && ecap == (uint64_t)-1) {
910 warn_invalid_dmar(drhd->address, " returns all ones");
911 return -EINVAL;
912 }
913
914 return 0;
915}
916
917void __init detect_intel_iommu(void)
918{
919 int ret;
920 struct dmar_res_callback validate_drhd_cb = {
921 .cb[ACPI_DMAR_TYPE_HARDWARE_UNIT] = &dmar_validate_one_drhd,
922 .ignore_unhandled = true,
923 };
924
925 down_write(&dmar_global_lock);
926 ret = dmar_table_detect();
927 if (!ret)
928 ret = dmar_walk_dmar_table((struct acpi_table_dmar *)dmar_tbl,
929 &validate_drhd_cb);
930 if (!ret && !no_iommu && !iommu_detected &&
931 (!dmar_disabled || dmar_platform_optin())) {
932 iommu_detected = 1;
933 /* Make sure ACS will be enabled */
934 pci_request_acs();
935 }
936
937#ifdef CONFIG_X86
938 if (!ret) {
939 x86_init.iommu.iommu_init = intel_iommu_init;
940 x86_platform.iommu_shutdown = intel_iommu_shutdown;
941 }
942
943#endif
944
945 if (dmar_tbl) {
946 acpi_put_table(dmar_tbl);
947 dmar_tbl = NULL;
948 }
949 up_write(&dmar_global_lock);
950}
951
952static void unmap_iommu(struct intel_iommu *iommu)
953{
954 iounmap(iommu->reg);
955 release_mem_region(iommu->reg_phys, iommu->reg_size);
956}
957
958/**
959 * map_iommu: map the iommu's registers
960 * @iommu: the iommu to map
961 * @drhd: DMA remapping hardware definition structure
962 *
963 * Memory map the iommu's registers. Start w/ a single page, and
964 * possibly expand if that turns out to be insufficent.
965 */
966static int map_iommu(struct intel_iommu *iommu, struct dmar_drhd_unit *drhd)
967{
968 u64 phys_addr = drhd->reg_base_addr;
969 int map_size, err=0;
970
971 iommu->reg_phys = phys_addr;
972 iommu->reg_size = drhd->reg_size;
973
974 if (!request_mem_region(iommu->reg_phys, iommu->reg_size, iommu->name)) {
975 pr_err("Can't reserve memory\n");
976 err = -EBUSY;
977 goto out;
978 }
979
980 iommu->reg = ioremap(iommu->reg_phys, iommu->reg_size);
981 if (!iommu->reg) {
982 pr_err("Can't map the region\n");
983 err = -ENOMEM;
984 goto release;
985 }
986
987 iommu->cap = dmar_readq(iommu->reg + DMAR_CAP_REG);
988 iommu->ecap = dmar_readq(iommu->reg + DMAR_ECAP_REG);
989
990 if (iommu->cap == (uint64_t)-1 && iommu->ecap == (uint64_t)-1) {
991 err = -EINVAL;
992 warn_invalid_dmar(phys_addr, " returns all ones");
993 goto unmap;
994 }
995
996 /* the registers might be more than one page */
997 map_size = max_t(int, ecap_max_iotlb_offset(iommu->ecap),
998 cap_max_fault_reg_offset(iommu->cap));
999 map_size = VTD_PAGE_ALIGN(map_size);
1000 if (map_size > iommu->reg_size) {
1001 iounmap(iommu->reg);
1002 release_mem_region(iommu->reg_phys, iommu->reg_size);
1003 iommu->reg_size = map_size;
1004 if (!request_mem_region(iommu->reg_phys, iommu->reg_size,
1005 iommu->name)) {
1006 pr_err("Can't reserve memory\n");
1007 err = -EBUSY;
1008 goto out;
1009 }
1010 iommu->reg = ioremap(iommu->reg_phys, iommu->reg_size);
1011 if (!iommu->reg) {
1012 pr_err("Can't map the region\n");
1013 err = -ENOMEM;
1014 goto release;
1015 }
1016 }
1017
1018 if (cap_ecmds(iommu->cap)) {
1019 int i;
1020
1021 for (i = 0; i < DMA_MAX_NUM_ECMDCAP; i++) {
1022 iommu->ecmdcap[i] = dmar_readq(iommu->reg + DMAR_ECCAP_REG +
1023 i * DMA_ECMD_REG_STEP);
1024 }
1025 }
1026
1027 err = 0;
1028 goto out;
1029
1030unmap:
1031 iounmap(iommu->reg);
1032release:
1033 release_mem_region(iommu->reg_phys, iommu->reg_size);
1034out:
1035 return err;
1036}
1037
1038static int alloc_iommu(struct dmar_drhd_unit *drhd)
1039{
1040 struct intel_iommu *iommu;
1041 u32 ver, sts;
1042 int agaw = -1;
1043 int msagaw = -1;
1044 int err;
1045
1046 if (!drhd->reg_base_addr) {
1047 warn_invalid_dmar(0, "");
1048 return -EINVAL;
1049 }
1050
1051 iommu = kzalloc(sizeof(*iommu), GFP_KERNEL);
1052 if (!iommu)
1053 return -ENOMEM;
1054
1055 iommu->seq_id = ida_alloc_range(&dmar_seq_ids, 0,
1056 DMAR_UNITS_SUPPORTED - 1, GFP_KERNEL);
1057 if (iommu->seq_id < 0) {
1058 pr_err("Failed to allocate seq_id\n");
1059 err = iommu->seq_id;
1060 goto error;
1061 }
1062 sprintf(iommu->name, "dmar%d", iommu->seq_id);
1063
1064 err = map_iommu(iommu, drhd);
1065 if (err) {
1066 pr_err("Failed to map %s\n", iommu->name);
1067 goto error_free_seq_id;
1068 }
1069
1070 err = -EINVAL;
1071 if (!cap_sagaw(iommu->cap) &&
1072 (!ecap_smts(iommu->ecap) || ecap_slts(iommu->ecap))) {
1073 pr_info("%s: No supported address widths. Not attempting DMA translation.\n",
1074 iommu->name);
1075 drhd->ignored = 1;
1076 }
1077
1078 if (!drhd->ignored) {
1079 agaw = iommu_calculate_agaw(iommu);
1080 if (agaw < 0) {
1081 pr_err("Cannot get a valid agaw for iommu (seq_id = %d)\n",
1082 iommu->seq_id);
1083 drhd->ignored = 1;
1084 }
1085 }
1086 if (!drhd->ignored) {
1087 msagaw = iommu_calculate_max_sagaw(iommu);
1088 if (msagaw < 0) {
1089 pr_err("Cannot get a valid max agaw for iommu (seq_id = %d)\n",
1090 iommu->seq_id);
1091 drhd->ignored = 1;
1092 agaw = -1;
1093 }
1094 }
1095 iommu->agaw = agaw;
1096 iommu->msagaw = msagaw;
1097 iommu->segment = drhd->segment;
1098
1099 iommu->node = NUMA_NO_NODE;
1100
1101 ver = readl(iommu->reg + DMAR_VER_REG);
1102 pr_info("%s: reg_base_addr %llx ver %d:%d cap %llx ecap %llx\n",
1103 iommu->name,
1104 (unsigned long long)drhd->reg_base_addr,
1105 DMAR_VER_MAJOR(ver), DMAR_VER_MINOR(ver),
1106 (unsigned long long)iommu->cap,
1107 (unsigned long long)iommu->ecap);
1108
1109 /* Reflect status in gcmd */
1110 sts = readl(iommu->reg + DMAR_GSTS_REG);
1111 if (sts & DMA_GSTS_IRES)
1112 iommu->gcmd |= DMA_GCMD_IRE;
1113 if (sts & DMA_GSTS_TES)
1114 iommu->gcmd |= DMA_GCMD_TE;
1115 if (sts & DMA_GSTS_QIES)
1116 iommu->gcmd |= DMA_GCMD_QIE;
1117
1118 if (alloc_iommu_pmu(iommu))
1119 pr_debug("Cannot alloc PMU for iommu (seq_id = %d)\n", iommu->seq_id);
1120
1121 raw_spin_lock_init(&iommu->register_lock);
1122
1123 /*
1124 * A value of N in PSS field of eCap register indicates hardware
1125 * supports PASID field of N+1 bits.
1126 */
1127 if (pasid_supported(iommu))
1128 iommu->iommu.max_pasids = 2UL << ecap_pss(iommu->ecap);
1129
1130 /*
1131 * This is only for hotplug; at boot time intel_iommu_enabled won't
1132 * be set yet. When intel_iommu_init() runs, it registers the units
1133 * present at boot time, then sets intel_iommu_enabled.
1134 */
1135 if (intel_iommu_enabled && !drhd->ignored) {
1136 err = iommu_device_sysfs_add(&iommu->iommu, NULL,
1137 intel_iommu_groups,
1138 "%s", iommu->name);
1139 if (err)
1140 goto err_unmap;
1141
1142 err = iommu_device_register(&iommu->iommu, &intel_iommu_ops, NULL);
1143 if (err)
1144 goto err_sysfs;
1145
1146 iommu_pmu_register(iommu);
1147 }
1148
1149 drhd->iommu = iommu;
1150 iommu->drhd = drhd;
1151
1152 return 0;
1153
1154err_sysfs:
1155 iommu_device_sysfs_remove(&iommu->iommu);
1156err_unmap:
1157 free_iommu_pmu(iommu);
1158 unmap_iommu(iommu);
1159error_free_seq_id:
1160 ida_free(&dmar_seq_ids, iommu->seq_id);
1161error:
1162 kfree(iommu);
1163 return err;
1164}
1165
1166static void free_iommu(struct intel_iommu *iommu)
1167{
1168 if (intel_iommu_enabled && !iommu->drhd->ignored) {
1169 iommu_pmu_unregister(iommu);
1170 iommu_device_unregister(&iommu->iommu);
1171 iommu_device_sysfs_remove(&iommu->iommu);
1172 }
1173
1174 free_iommu_pmu(iommu);
1175
1176 if (iommu->irq) {
1177 if (iommu->pr_irq) {
1178 free_irq(iommu->pr_irq, iommu);
1179 dmar_free_hwirq(iommu->pr_irq);
1180 iommu->pr_irq = 0;
1181 }
1182 free_irq(iommu->irq, iommu);
1183 dmar_free_hwirq(iommu->irq);
1184 iommu->irq = 0;
1185 }
1186
1187 if (iommu->qi) {
1188 free_page((unsigned long)iommu->qi->desc);
1189 kfree(iommu->qi->desc_status);
1190 kfree(iommu->qi);
1191 }
1192
1193 if (iommu->reg)
1194 unmap_iommu(iommu);
1195
1196 ida_free(&dmar_seq_ids, iommu->seq_id);
1197 kfree(iommu);
1198}
1199
1200/*
1201 * Reclaim all the submitted descriptors which have completed its work.
1202 */
1203static inline void reclaim_free_desc(struct q_inval *qi)
1204{
1205 while (qi->desc_status[qi->free_tail] == QI_DONE ||
1206 qi->desc_status[qi->free_tail] == QI_ABORT) {
1207 qi->desc_status[qi->free_tail] = QI_FREE;
1208 qi->free_tail = (qi->free_tail + 1) % QI_LENGTH;
1209 qi->free_cnt++;
1210 }
1211}
1212
1213static const char *qi_type_string(u8 type)
1214{
1215 switch (type) {
1216 case QI_CC_TYPE:
1217 return "Context-cache Invalidation";
1218 case QI_IOTLB_TYPE:
1219 return "IOTLB Invalidation";
1220 case QI_DIOTLB_TYPE:
1221 return "Device-TLB Invalidation";
1222 case QI_IEC_TYPE:
1223 return "Interrupt Entry Cache Invalidation";
1224 case QI_IWD_TYPE:
1225 return "Invalidation Wait";
1226 case QI_EIOTLB_TYPE:
1227 return "PASID-based IOTLB Invalidation";
1228 case QI_PC_TYPE:
1229 return "PASID-cache Invalidation";
1230 case QI_DEIOTLB_TYPE:
1231 return "PASID-based Device-TLB Invalidation";
1232 case QI_PGRP_RESP_TYPE:
1233 return "Page Group Response";
1234 default:
1235 return "UNKNOWN";
1236 }
1237}
1238
1239static void qi_dump_fault(struct intel_iommu *iommu, u32 fault)
1240{
1241 unsigned int head = dmar_readl(iommu->reg + DMAR_IQH_REG);
1242 u64 iqe_err = dmar_readq(iommu->reg + DMAR_IQER_REG);
1243 struct qi_desc *desc = iommu->qi->desc + head;
1244
1245 if (fault & DMA_FSTS_IQE)
1246 pr_err("VT-d detected Invalidation Queue Error: Reason %llx",
1247 DMAR_IQER_REG_IQEI(iqe_err));
1248 if (fault & DMA_FSTS_ITE)
1249 pr_err("VT-d detected Invalidation Time-out Error: SID %llx",
1250 DMAR_IQER_REG_ITESID(iqe_err));
1251 if (fault & DMA_FSTS_ICE)
1252 pr_err("VT-d detected Invalidation Completion Error: SID %llx",
1253 DMAR_IQER_REG_ICESID(iqe_err));
1254
1255 pr_err("QI HEAD: %s qw0 = 0x%llx, qw1 = 0x%llx\n",
1256 qi_type_string(desc->qw0 & 0xf),
1257 (unsigned long long)desc->qw0,
1258 (unsigned long long)desc->qw1);
1259
1260 head = ((head >> qi_shift(iommu)) + QI_LENGTH - 1) % QI_LENGTH;
1261 head <<= qi_shift(iommu);
1262 desc = iommu->qi->desc + head;
1263
1264 pr_err("QI PRIOR: %s qw0 = 0x%llx, qw1 = 0x%llx\n",
1265 qi_type_string(desc->qw0 & 0xf),
1266 (unsigned long long)desc->qw0,
1267 (unsigned long long)desc->qw1);
1268}
1269
1270static int qi_check_fault(struct intel_iommu *iommu, int index, int wait_index)
1271{
1272 u32 fault;
1273 int head, tail;
1274 struct q_inval *qi = iommu->qi;
1275 int shift = qi_shift(iommu);
1276
1277 if (qi->desc_status[wait_index] == QI_ABORT)
1278 return -EAGAIN;
1279
1280 fault = readl(iommu->reg + DMAR_FSTS_REG);
1281 if (fault & (DMA_FSTS_IQE | DMA_FSTS_ITE | DMA_FSTS_ICE))
1282 qi_dump_fault(iommu, fault);
1283
1284 /*
1285 * If IQE happens, the head points to the descriptor associated
1286 * with the error. No new descriptors are fetched until the IQE
1287 * is cleared.
1288 */
1289 if (fault & DMA_FSTS_IQE) {
1290 head = readl(iommu->reg + DMAR_IQH_REG);
1291 if ((head >> shift) == index) {
1292 struct qi_desc *desc = qi->desc + head;
1293
1294 /*
1295 * desc->qw2 and desc->qw3 are either reserved or
1296 * used by software as private data. We won't print
1297 * out these two qw's for security consideration.
1298 */
1299 memcpy(desc, qi->desc + (wait_index << shift),
1300 1 << shift);
1301 writel(DMA_FSTS_IQE, iommu->reg + DMAR_FSTS_REG);
1302 pr_info("Invalidation Queue Error (IQE) cleared\n");
1303 return -EINVAL;
1304 }
1305 }
1306
1307 /*
1308 * If ITE happens, all pending wait_desc commands are aborted.
1309 * No new descriptors are fetched until the ITE is cleared.
1310 */
1311 if (fault & DMA_FSTS_ITE) {
1312 head = readl(iommu->reg + DMAR_IQH_REG);
1313 head = ((head >> shift) - 1 + QI_LENGTH) % QI_LENGTH;
1314 head |= 1;
1315 tail = readl(iommu->reg + DMAR_IQT_REG);
1316 tail = ((tail >> shift) - 1 + QI_LENGTH) % QI_LENGTH;
1317
1318 writel(DMA_FSTS_ITE, iommu->reg + DMAR_FSTS_REG);
1319 pr_info("Invalidation Time-out Error (ITE) cleared\n");
1320
1321 do {
1322 if (qi->desc_status[head] == QI_IN_USE)
1323 qi->desc_status[head] = QI_ABORT;
1324 head = (head - 2 + QI_LENGTH) % QI_LENGTH;
1325 } while (head != tail);
1326
1327 if (qi->desc_status[wait_index] == QI_ABORT)
1328 return -EAGAIN;
1329 }
1330
1331 if (fault & DMA_FSTS_ICE) {
1332 writel(DMA_FSTS_ICE, iommu->reg + DMAR_FSTS_REG);
1333 pr_info("Invalidation Completion Error (ICE) cleared\n");
1334 }
1335
1336 return 0;
1337}
1338
1339/*
1340 * Function to submit invalidation descriptors of all types to the queued
1341 * invalidation interface(QI). Multiple descriptors can be submitted at a
1342 * time, a wait descriptor will be appended to each submission to ensure
1343 * hardware has completed the invalidation before return. Wait descriptors
1344 * can be part of the submission but it will not be polled for completion.
1345 */
1346int qi_submit_sync(struct intel_iommu *iommu, struct qi_desc *desc,
1347 unsigned int count, unsigned long options)
1348{
1349 struct q_inval *qi = iommu->qi;
1350 s64 devtlb_start_ktime = 0;
1351 s64 iotlb_start_ktime = 0;
1352 s64 iec_start_ktime = 0;
1353 struct qi_desc wait_desc;
1354 int wait_index, index;
1355 unsigned long flags;
1356 int offset, shift;
1357 int rc, i;
1358 u64 type;
1359
1360 if (!qi)
1361 return 0;
1362
1363 type = desc->qw0 & GENMASK_ULL(3, 0);
1364
1365 if ((type == QI_IOTLB_TYPE || type == QI_EIOTLB_TYPE) &&
1366 dmar_latency_enabled(iommu, DMAR_LATENCY_INV_IOTLB))
1367 iotlb_start_ktime = ktime_to_ns(ktime_get());
1368
1369 if ((type == QI_DIOTLB_TYPE || type == QI_DEIOTLB_TYPE) &&
1370 dmar_latency_enabled(iommu, DMAR_LATENCY_INV_DEVTLB))
1371 devtlb_start_ktime = ktime_to_ns(ktime_get());
1372
1373 if (type == QI_IEC_TYPE &&
1374 dmar_latency_enabled(iommu, DMAR_LATENCY_INV_IEC))
1375 iec_start_ktime = ktime_to_ns(ktime_get());
1376
1377restart:
1378 rc = 0;
1379
1380 raw_spin_lock_irqsave(&qi->q_lock, flags);
1381 /*
1382 * Check if we have enough empty slots in the queue to submit,
1383 * the calculation is based on:
1384 * # of desc + 1 wait desc + 1 space between head and tail
1385 */
1386 while (qi->free_cnt < count + 2) {
1387 raw_spin_unlock_irqrestore(&qi->q_lock, flags);
1388 cpu_relax();
1389 raw_spin_lock_irqsave(&qi->q_lock, flags);
1390 }
1391
1392 index = qi->free_head;
1393 wait_index = (index + count) % QI_LENGTH;
1394 shift = qi_shift(iommu);
1395
1396 for (i = 0; i < count; i++) {
1397 offset = ((index + i) % QI_LENGTH) << shift;
1398 memcpy(qi->desc + offset, &desc[i], 1 << shift);
1399 qi->desc_status[(index + i) % QI_LENGTH] = QI_IN_USE;
1400 trace_qi_submit(iommu, desc[i].qw0, desc[i].qw1,
1401 desc[i].qw2, desc[i].qw3);
1402 }
1403 qi->desc_status[wait_index] = QI_IN_USE;
1404
1405 wait_desc.qw0 = QI_IWD_STATUS_DATA(QI_DONE) |
1406 QI_IWD_STATUS_WRITE | QI_IWD_TYPE;
1407 if (options & QI_OPT_WAIT_DRAIN)
1408 wait_desc.qw0 |= QI_IWD_PRQ_DRAIN;
1409 wait_desc.qw1 = virt_to_phys(&qi->desc_status[wait_index]);
1410 wait_desc.qw2 = 0;
1411 wait_desc.qw3 = 0;
1412
1413 offset = wait_index << shift;
1414 memcpy(qi->desc + offset, &wait_desc, 1 << shift);
1415
1416 qi->free_head = (qi->free_head + count + 1) % QI_LENGTH;
1417 qi->free_cnt -= count + 1;
1418
1419 /*
1420 * update the HW tail register indicating the presence of
1421 * new descriptors.
1422 */
1423 writel(qi->free_head << shift, iommu->reg + DMAR_IQT_REG);
1424
1425 while (qi->desc_status[wait_index] != QI_DONE) {
1426 /*
1427 * We will leave the interrupts disabled, to prevent interrupt
1428 * context to queue another cmd while a cmd is already submitted
1429 * and waiting for completion on this cpu. This is to avoid
1430 * a deadlock where the interrupt context can wait indefinitely
1431 * for free slots in the queue.
1432 */
1433 rc = qi_check_fault(iommu, index, wait_index);
1434 if (rc)
1435 break;
1436
1437 raw_spin_unlock(&qi->q_lock);
1438 cpu_relax();
1439 raw_spin_lock(&qi->q_lock);
1440 }
1441
1442 for (i = 0; i < count; i++)
1443 qi->desc_status[(index + i) % QI_LENGTH] = QI_DONE;
1444
1445 reclaim_free_desc(qi);
1446 raw_spin_unlock_irqrestore(&qi->q_lock, flags);
1447
1448 if (rc == -EAGAIN)
1449 goto restart;
1450
1451 if (iotlb_start_ktime)
1452 dmar_latency_update(iommu, DMAR_LATENCY_INV_IOTLB,
1453 ktime_to_ns(ktime_get()) - iotlb_start_ktime);
1454
1455 if (devtlb_start_ktime)
1456 dmar_latency_update(iommu, DMAR_LATENCY_INV_DEVTLB,
1457 ktime_to_ns(ktime_get()) - devtlb_start_ktime);
1458
1459 if (iec_start_ktime)
1460 dmar_latency_update(iommu, DMAR_LATENCY_INV_IEC,
1461 ktime_to_ns(ktime_get()) - iec_start_ktime);
1462
1463 return rc;
1464}
1465
1466/*
1467 * Flush the global interrupt entry cache.
1468 */
1469void qi_global_iec(struct intel_iommu *iommu)
1470{
1471 struct qi_desc desc;
1472
1473 desc.qw0 = QI_IEC_TYPE;
1474 desc.qw1 = 0;
1475 desc.qw2 = 0;
1476 desc.qw3 = 0;
1477
1478 /* should never fail */
1479 qi_submit_sync(iommu, &desc, 1, 0);
1480}
1481
1482void qi_flush_context(struct intel_iommu *iommu, u16 did, u16 sid, u8 fm,
1483 u64 type)
1484{
1485 struct qi_desc desc;
1486
1487 desc.qw0 = QI_CC_FM(fm) | QI_CC_SID(sid) | QI_CC_DID(did)
1488 | QI_CC_GRAN(type) | QI_CC_TYPE;
1489 desc.qw1 = 0;
1490 desc.qw2 = 0;
1491 desc.qw3 = 0;
1492
1493 qi_submit_sync(iommu, &desc, 1, 0);
1494}
1495
1496void qi_flush_iotlb(struct intel_iommu *iommu, u16 did, u64 addr,
1497 unsigned int size_order, u64 type)
1498{
1499 u8 dw = 0, dr = 0;
1500
1501 struct qi_desc desc;
1502 int ih = 0;
1503
1504 if (cap_write_drain(iommu->cap))
1505 dw = 1;
1506
1507 if (cap_read_drain(iommu->cap))
1508 dr = 1;
1509
1510 desc.qw0 = QI_IOTLB_DID(did) | QI_IOTLB_DR(dr) | QI_IOTLB_DW(dw)
1511 | QI_IOTLB_GRAN(type) | QI_IOTLB_TYPE;
1512 desc.qw1 = QI_IOTLB_ADDR(addr) | QI_IOTLB_IH(ih)
1513 | QI_IOTLB_AM(size_order);
1514 desc.qw2 = 0;
1515 desc.qw3 = 0;
1516
1517 qi_submit_sync(iommu, &desc, 1, 0);
1518}
1519
1520void qi_flush_dev_iotlb(struct intel_iommu *iommu, u16 sid, u16 pfsid,
1521 u16 qdep, u64 addr, unsigned mask)
1522{
1523 struct qi_desc desc;
1524
1525 /*
1526 * VT-d spec, section 4.3:
1527 *
1528 * Software is recommended to not submit any Device-TLB invalidation
1529 * requests while address remapping hardware is disabled.
1530 */
1531 if (!(iommu->gcmd & DMA_GCMD_TE))
1532 return;
1533
1534 if (mask) {
1535 addr |= (1ULL << (VTD_PAGE_SHIFT + mask - 1)) - 1;
1536 desc.qw1 = QI_DEV_IOTLB_ADDR(addr) | QI_DEV_IOTLB_SIZE;
1537 } else
1538 desc.qw1 = QI_DEV_IOTLB_ADDR(addr);
1539
1540 if (qdep >= QI_DEV_IOTLB_MAX_INVS)
1541 qdep = 0;
1542
1543 desc.qw0 = QI_DEV_IOTLB_SID(sid) | QI_DEV_IOTLB_QDEP(qdep) |
1544 QI_DIOTLB_TYPE | QI_DEV_IOTLB_PFSID(pfsid);
1545 desc.qw2 = 0;
1546 desc.qw3 = 0;
1547
1548 qi_submit_sync(iommu, &desc, 1, 0);
1549}
1550
1551/* PASID-based IOTLB invalidation */
1552void qi_flush_piotlb(struct intel_iommu *iommu, u16 did, u32 pasid, u64 addr,
1553 unsigned long npages, bool ih)
1554{
1555 struct qi_desc desc = {.qw2 = 0, .qw3 = 0};
1556
1557 /*
1558 * npages == -1 means a PASID-selective invalidation, otherwise,
1559 * a positive value for Page-selective-within-PASID invalidation.
1560 * 0 is not a valid input.
1561 */
1562 if (WARN_ON(!npages)) {
1563 pr_err("Invalid input npages = %ld\n", npages);
1564 return;
1565 }
1566
1567 if (npages == -1) {
1568 desc.qw0 = QI_EIOTLB_PASID(pasid) |
1569 QI_EIOTLB_DID(did) |
1570 QI_EIOTLB_GRAN(QI_GRAN_NONG_PASID) |
1571 QI_EIOTLB_TYPE;
1572 desc.qw1 = 0;
1573 } else {
1574 int mask = ilog2(__roundup_pow_of_two(npages));
1575 unsigned long align = (1ULL << (VTD_PAGE_SHIFT + mask));
1576
1577 if (WARN_ON_ONCE(!IS_ALIGNED(addr, align)))
1578 addr = ALIGN_DOWN(addr, align);
1579
1580 desc.qw0 = QI_EIOTLB_PASID(pasid) |
1581 QI_EIOTLB_DID(did) |
1582 QI_EIOTLB_GRAN(QI_GRAN_PSI_PASID) |
1583 QI_EIOTLB_TYPE;
1584 desc.qw1 = QI_EIOTLB_ADDR(addr) |
1585 QI_EIOTLB_IH(ih) |
1586 QI_EIOTLB_AM(mask);
1587 }
1588
1589 qi_submit_sync(iommu, &desc, 1, 0);
1590}
1591
1592/* PASID-based device IOTLB Invalidate */
1593void qi_flush_dev_iotlb_pasid(struct intel_iommu *iommu, u16 sid, u16 pfsid,
1594 u32 pasid, u16 qdep, u64 addr, unsigned int size_order)
1595{
1596 unsigned long mask = 1UL << (VTD_PAGE_SHIFT + size_order - 1);
1597 struct qi_desc desc = {.qw1 = 0, .qw2 = 0, .qw3 = 0};
1598
1599 /*
1600 * VT-d spec, section 4.3:
1601 *
1602 * Software is recommended to not submit any Device-TLB invalidation
1603 * requests while address remapping hardware is disabled.
1604 */
1605 if (!(iommu->gcmd & DMA_GCMD_TE))
1606 return;
1607
1608 desc.qw0 = QI_DEV_EIOTLB_PASID(pasid) | QI_DEV_EIOTLB_SID(sid) |
1609 QI_DEV_EIOTLB_QDEP(qdep) | QI_DEIOTLB_TYPE |
1610 QI_DEV_IOTLB_PFSID(pfsid);
1611
1612 /*
1613 * If S bit is 0, we only flush a single page. If S bit is set,
1614 * The least significant zero bit indicates the invalidation address
1615 * range. VT-d spec 6.5.2.6.
1616 * e.g. address bit 12[0] indicates 8KB, 13[0] indicates 16KB.
1617 * size order = 0 is PAGE_SIZE 4KB
1618 * Max Invs Pending (MIP) is set to 0 for now until we have DIT in
1619 * ECAP.
1620 */
1621 if (!IS_ALIGNED(addr, VTD_PAGE_SIZE << size_order))
1622 pr_warn_ratelimited("Invalidate non-aligned address %llx, order %d\n",
1623 addr, size_order);
1624
1625 /* Take page address */
1626 desc.qw1 = QI_DEV_EIOTLB_ADDR(addr);
1627
1628 if (size_order) {
1629 /*
1630 * Existing 0s in address below size_order may be the least
1631 * significant bit, we must set them to 1s to avoid having
1632 * smaller size than desired.
1633 */
1634 desc.qw1 |= GENMASK_ULL(size_order + VTD_PAGE_SHIFT - 1,
1635 VTD_PAGE_SHIFT);
1636 /* Clear size_order bit to indicate size */
1637 desc.qw1 &= ~mask;
1638 /* Set the S bit to indicate flushing more than 1 page */
1639 desc.qw1 |= QI_DEV_EIOTLB_SIZE;
1640 }
1641
1642 qi_submit_sync(iommu, &desc, 1, 0);
1643}
1644
1645void qi_flush_pasid_cache(struct intel_iommu *iommu, u16 did,
1646 u64 granu, u32 pasid)
1647{
1648 struct qi_desc desc = {.qw1 = 0, .qw2 = 0, .qw3 = 0};
1649
1650 desc.qw0 = QI_PC_PASID(pasid) | QI_PC_DID(did) |
1651 QI_PC_GRAN(granu) | QI_PC_TYPE;
1652 qi_submit_sync(iommu, &desc, 1, 0);
1653}
1654
1655/*
1656 * Disable Queued Invalidation interface.
1657 */
1658void dmar_disable_qi(struct intel_iommu *iommu)
1659{
1660 unsigned long flags;
1661 u32 sts;
1662 cycles_t start_time = get_cycles();
1663
1664 if (!ecap_qis(iommu->ecap))
1665 return;
1666
1667 raw_spin_lock_irqsave(&iommu->register_lock, flags);
1668
1669 sts = readl(iommu->reg + DMAR_GSTS_REG);
1670 if (!(sts & DMA_GSTS_QIES))
1671 goto end;
1672
1673 /*
1674 * Give a chance to HW to complete the pending invalidation requests.
1675 */
1676 while ((readl(iommu->reg + DMAR_IQT_REG) !=
1677 readl(iommu->reg + DMAR_IQH_REG)) &&
1678 (DMAR_OPERATION_TIMEOUT > (get_cycles() - start_time)))
1679 cpu_relax();
1680
1681 iommu->gcmd &= ~DMA_GCMD_QIE;
1682 writel(iommu->gcmd, iommu->reg + DMAR_GCMD_REG);
1683
1684 IOMMU_WAIT_OP(iommu, DMAR_GSTS_REG, readl,
1685 !(sts & DMA_GSTS_QIES), sts);
1686end:
1687 raw_spin_unlock_irqrestore(&iommu->register_lock, flags);
1688}
1689
1690/*
1691 * Enable queued invalidation.
1692 */
1693static void __dmar_enable_qi(struct intel_iommu *iommu)
1694{
1695 u32 sts;
1696 unsigned long flags;
1697 struct q_inval *qi = iommu->qi;
1698 u64 val = virt_to_phys(qi->desc);
1699
1700 qi->free_head = qi->free_tail = 0;
1701 qi->free_cnt = QI_LENGTH;
1702
1703 /*
1704 * Set DW=1 and QS=1 in IQA_REG when Scalable Mode capability
1705 * is present.
1706 */
1707 if (ecap_smts(iommu->ecap))
1708 val |= BIT_ULL(11) | BIT_ULL(0);
1709
1710 raw_spin_lock_irqsave(&iommu->register_lock, flags);
1711
1712 /* write zero to the tail reg */
1713 writel(0, iommu->reg + DMAR_IQT_REG);
1714
1715 dmar_writeq(iommu->reg + DMAR_IQA_REG, val);
1716
1717 iommu->gcmd |= DMA_GCMD_QIE;
1718 writel(iommu->gcmd, iommu->reg + DMAR_GCMD_REG);
1719
1720 /* Make sure hardware complete it */
1721 IOMMU_WAIT_OP(iommu, DMAR_GSTS_REG, readl, (sts & DMA_GSTS_QIES), sts);
1722
1723 raw_spin_unlock_irqrestore(&iommu->register_lock, flags);
1724}
1725
1726/*
1727 * Enable Queued Invalidation interface. This is a must to support
1728 * interrupt-remapping. Also used by DMA-remapping, which replaces
1729 * register based IOTLB invalidation.
1730 */
1731int dmar_enable_qi(struct intel_iommu *iommu)
1732{
1733 struct q_inval *qi;
1734 struct page *desc_page;
1735
1736 if (!ecap_qis(iommu->ecap))
1737 return -ENOENT;
1738
1739 /*
1740 * queued invalidation is already setup and enabled.
1741 */
1742 if (iommu->qi)
1743 return 0;
1744
1745 iommu->qi = kmalloc(sizeof(*qi), GFP_ATOMIC);
1746 if (!iommu->qi)
1747 return -ENOMEM;
1748
1749 qi = iommu->qi;
1750
1751 /*
1752 * Need two pages to accommodate 256 descriptors of 256 bits each
1753 * if the remapping hardware supports scalable mode translation.
1754 */
1755 desc_page = alloc_pages_node(iommu->node, GFP_ATOMIC | __GFP_ZERO,
1756 !!ecap_smts(iommu->ecap));
1757 if (!desc_page) {
1758 kfree(qi);
1759 iommu->qi = NULL;
1760 return -ENOMEM;
1761 }
1762
1763 qi->desc = page_address(desc_page);
1764
1765 qi->desc_status = kcalloc(QI_LENGTH, sizeof(int), GFP_ATOMIC);
1766 if (!qi->desc_status) {
1767 free_page((unsigned long) qi->desc);
1768 kfree(qi);
1769 iommu->qi = NULL;
1770 return -ENOMEM;
1771 }
1772
1773 raw_spin_lock_init(&qi->q_lock);
1774
1775 __dmar_enable_qi(iommu);
1776
1777 return 0;
1778}
1779
1780/* iommu interrupt handling. Most stuff are MSI-like. */
1781
1782enum faulttype {
1783 DMA_REMAP,
1784 INTR_REMAP,
1785 UNKNOWN,
1786};
1787
1788static const char *dma_remap_fault_reasons[] =
1789{
1790 "Software",
1791 "Present bit in root entry is clear",
1792 "Present bit in context entry is clear",
1793 "Invalid context entry",
1794 "Access beyond MGAW",
1795 "PTE Write access is not set",
1796 "PTE Read access is not set",
1797 "Next page table ptr is invalid",
1798 "Root table address invalid",
1799 "Context table ptr is invalid",
1800 "non-zero reserved fields in RTP",
1801 "non-zero reserved fields in CTP",
1802 "non-zero reserved fields in PTE",
1803 "PCE for translation request specifies blocking",
1804};
1805
1806static const char * const dma_remap_sm_fault_reasons[] = {
1807 "SM: Invalid Root Table Address",
1808 "SM: TTM 0 for request with PASID",
1809 "SM: TTM 0 for page group request",
1810 "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", /* 0x33-0x37 */
1811 "SM: Error attempting to access Root Entry",
1812 "SM: Present bit in Root Entry is clear",
1813 "SM: Non-zero reserved field set in Root Entry",
1814 "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", /* 0x3B-0x3F */
1815 "SM: Error attempting to access Context Entry",
1816 "SM: Present bit in Context Entry is clear",
1817 "SM: Non-zero reserved field set in the Context Entry",
1818 "SM: Invalid Context Entry",
1819 "SM: DTE field in Context Entry is clear",
1820 "SM: PASID Enable field in Context Entry is clear",
1821 "SM: PASID is larger than the max in Context Entry",
1822 "SM: PRE field in Context-Entry is clear",
1823 "SM: RID_PASID field error in Context-Entry",
1824 "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", /* 0x49-0x4F */
1825 "SM: Error attempting to access the PASID Directory Entry",
1826 "SM: Present bit in Directory Entry is clear",
1827 "SM: Non-zero reserved field set in PASID Directory Entry",
1828 "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", /* 0x53-0x57 */
1829 "SM: Error attempting to access PASID Table Entry",
1830 "SM: Present bit in PASID Table Entry is clear",
1831 "SM: Non-zero reserved field set in PASID Table Entry",
1832 "SM: Invalid Scalable-Mode PASID Table Entry",
1833 "SM: ERE field is clear in PASID Table Entry",
1834 "SM: SRE field is clear in PASID Table Entry",
1835 "Unknown", "Unknown",/* 0x5E-0x5F */
1836 "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", /* 0x60-0x67 */
1837 "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", /* 0x68-0x6F */
1838 "SM: Error attempting to access first-level paging entry",
1839 "SM: Present bit in first-level paging entry is clear",
1840 "SM: Non-zero reserved field set in first-level paging entry",
1841 "SM: Error attempting to access FL-PML4 entry",
1842 "SM: First-level entry address beyond MGAW in Nested translation",
1843 "SM: Read permission error in FL-PML4 entry in Nested translation",
1844 "SM: Read permission error in first-level paging entry in Nested translation",
1845 "SM: Write permission error in first-level paging entry in Nested translation",
1846 "SM: Error attempting to access second-level paging entry",
1847 "SM: Read/Write permission error in second-level paging entry",
1848 "SM: Non-zero reserved field set in second-level paging entry",
1849 "SM: Invalid second-level page table pointer",
1850 "SM: A/D bit update needed in second-level entry when set up in no snoop",
1851 "Unknown", "Unknown", "Unknown", /* 0x7D-0x7F */
1852 "SM: Address in first-level translation is not canonical",
1853 "SM: U/S set 0 for first-level translation with user privilege",
1854 "SM: No execute permission for request with PASID and ER=1",
1855 "SM: Address beyond the DMA hardware max",
1856 "SM: Second-level entry address beyond the max",
1857 "SM: No write permission for Write/AtomicOp request",
1858 "SM: No read permission for Read/AtomicOp request",
1859 "SM: Invalid address-interrupt address",
1860 "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", "Unknown", /* 0x88-0x8F */
1861 "SM: A/D bit update needed in first-level entry when set up in no snoop",
1862};
1863
1864static const char *irq_remap_fault_reasons[] =
1865{
1866 "Detected reserved fields in the decoded interrupt-remapped request",
1867 "Interrupt index exceeded the interrupt-remapping table size",
1868 "Present field in the IRTE entry is clear",
1869 "Error accessing interrupt-remapping table pointed by IRTA_REG",
1870 "Detected reserved fields in the IRTE entry",
1871 "Blocked a compatibility format interrupt request",
1872 "Blocked an interrupt request due to source-id verification failure",
1873};
1874
1875static const char *dmar_get_fault_reason(u8 fault_reason, int *fault_type)
1876{
1877 if (fault_reason >= 0x20 && (fault_reason - 0x20 <
1878 ARRAY_SIZE(irq_remap_fault_reasons))) {
1879 *fault_type = INTR_REMAP;
1880 return irq_remap_fault_reasons[fault_reason - 0x20];
1881 } else if (fault_reason >= 0x30 && (fault_reason - 0x30 <
1882 ARRAY_SIZE(dma_remap_sm_fault_reasons))) {
1883 *fault_type = DMA_REMAP;
1884 return dma_remap_sm_fault_reasons[fault_reason - 0x30];
1885 } else if (fault_reason < ARRAY_SIZE(dma_remap_fault_reasons)) {
1886 *fault_type = DMA_REMAP;
1887 return dma_remap_fault_reasons[fault_reason];
1888 } else {
1889 *fault_type = UNKNOWN;
1890 return "Unknown";
1891 }
1892}
1893
1894
1895static inline int dmar_msi_reg(struct intel_iommu *iommu, int irq)
1896{
1897 if (iommu->irq == irq)
1898 return DMAR_FECTL_REG;
1899 else if (iommu->pr_irq == irq)
1900 return DMAR_PECTL_REG;
1901 else if (iommu->perf_irq == irq)
1902 return DMAR_PERFINTRCTL_REG;
1903 else
1904 BUG();
1905}
1906
1907void dmar_msi_unmask(struct irq_data *data)
1908{
1909 struct intel_iommu *iommu = irq_data_get_irq_handler_data(data);
1910 int reg = dmar_msi_reg(iommu, data->irq);
1911 unsigned long flag;
1912
1913 /* unmask it */
1914 raw_spin_lock_irqsave(&iommu->register_lock, flag);
1915 writel(0, iommu->reg + reg);
1916 /* Read a reg to force flush the post write */
1917 readl(iommu->reg + reg);
1918 raw_spin_unlock_irqrestore(&iommu->register_lock, flag);
1919}
1920
1921void dmar_msi_mask(struct irq_data *data)
1922{
1923 struct intel_iommu *iommu = irq_data_get_irq_handler_data(data);
1924 int reg = dmar_msi_reg(iommu, data->irq);
1925 unsigned long flag;
1926
1927 /* mask it */
1928 raw_spin_lock_irqsave(&iommu->register_lock, flag);
1929 writel(DMA_FECTL_IM, iommu->reg + reg);
1930 /* Read a reg to force flush the post write */
1931 readl(iommu->reg + reg);
1932 raw_spin_unlock_irqrestore(&iommu->register_lock, flag);
1933}
1934
1935void dmar_msi_write(int irq, struct msi_msg *msg)
1936{
1937 struct intel_iommu *iommu = irq_get_handler_data(irq);
1938 int reg = dmar_msi_reg(iommu, irq);
1939 unsigned long flag;
1940
1941 raw_spin_lock_irqsave(&iommu->register_lock, flag);
1942 writel(msg->data, iommu->reg + reg + 4);
1943 writel(msg->address_lo, iommu->reg + reg + 8);
1944 writel(msg->address_hi, iommu->reg + reg + 12);
1945 raw_spin_unlock_irqrestore(&iommu->register_lock, flag);
1946}
1947
1948void dmar_msi_read(int irq, struct msi_msg *msg)
1949{
1950 struct intel_iommu *iommu = irq_get_handler_data(irq);
1951 int reg = dmar_msi_reg(iommu, irq);
1952 unsigned long flag;
1953
1954 raw_spin_lock_irqsave(&iommu->register_lock, flag);
1955 msg->data = readl(iommu->reg + reg + 4);
1956 msg->address_lo = readl(iommu->reg + reg + 8);
1957 msg->address_hi = readl(iommu->reg + reg + 12);
1958 raw_spin_unlock_irqrestore(&iommu->register_lock, flag);
1959}
1960
1961static int dmar_fault_do_one(struct intel_iommu *iommu, int type,
1962 u8 fault_reason, u32 pasid, u16 source_id,
1963 unsigned long long addr)
1964{
1965 const char *reason;
1966 int fault_type;
1967
1968 reason = dmar_get_fault_reason(fault_reason, &fault_type);
1969
1970 if (fault_type == INTR_REMAP) {
1971 pr_err("[INTR-REMAP] Request device [%02x:%02x.%d] fault index 0x%llx [fault reason 0x%02x] %s\n",
1972 source_id >> 8, PCI_SLOT(source_id & 0xFF),
1973 PCI_FUNC(source_id & 0xFF), addr >> 48,
1974 fault_reason, reason);
1975
1976 return 0;
1977 }
1978
1979 if (pasid == IOMMU_PASID_INVALID)
1980 pr_err("[%s NO_PASID] Request device [%02x:%02x.%d] fault addr 0x%llx [fault reason 0x%02x] %s\n",
1981 type ? "DMA Read" : "DMA Write",
1982 source_id >> 8, PCI_SLOT(source_id & 0xFF),
1983 PCI_FUNC(source_id & 0xFF), addr,
1984 fault_reason, reason);
1985 else
1986 pr_err("[%s PASID 0x%x] Request device [%02x:%02x.%d] fault addr 0x%llx [fault reason 0x%02x] %s\n",
1987 type ? "DMA Read" : "DMA Write", pasid,
1988 source_id >> 8, PCI_SLOT(source_id & 0xFF),
1989 PCI_FUNC(source_id & 0xFF), addr,
1990 fault_reason, reason);
1991
1992 dmar_fault_dump_ptes(iommu, source_id, addr, pasid);
1993
1994 return 0;
1995}
1996
1997#define PRIMARY_FAULT_REG_LEN (16)
1998irqreturn_t dmar_fault(int irq, void *dev_id)
1999{
2000 struct intel_iommu *iommu = dev_id;
2001 int reg, fault_index;
2002 u32 fault_status;
2003 unsigned long flag;
2004 static DEFINE_RATELIMIT_STATE(rs,
2005 DEFAULT_RATELIMIT_INTERVAL,
2006 DEFAULT_RATELIMIT_BURST);
2007
2008 raw_spin_lock_irqsave(&iommu->register_lock, flag);
2009 fault_status = readl(iommu->reg + DMAR_FSTS_REG);
2010 if (fault_status && __ratelimit(&rs))
2011 pr_err("DRHD: handling fault status reg %x\n", fault_status);
2012
2013 /* TBD: ignore advanced fault log currently */
2014 if (!(fault_status & DMA_FSTS_PPF))
2015 goto unlock_exit;
2016
2017 fault_index = dma_fsts_fault_record_index(fault_status);
2018 reg = cap_fault_reg_offset(iommu->cap);
2019 while (1) {
2020 /* Disable printing, simply clear the fault when ratelimited */
2021 bool ratelimited = !__ratelimit(&rs);
2022 u8 fault_reason;
2023 u16 source_id;
2024 u64 guest_addr;
2025 u32 pasid;
2026 int type;
2027 u32 data;
2028 bool pasid_present;
2029
2030 /* highest 32 bits */
2031 data = readl(iommu->reg + reg +
2032 fault_index * PRIMARY_FAULT_REG_LEN + 12);
2033 if (!(data & DMA_FRCD_F))
2034 break;
2035
2036 if (!ratelimited) {
2037 fault_reason = dma_frcd_fault_reason(data);
2038 type = dma_frcd_type(data);
2039
2040 pasid = dma_frcd_pasid_value(data);
2041 data = readl(iommu->reg + reg +
2042 fault_index * PRIMARY_FAULT_REG_LEN + 8);
2043 source_id = dma_frcd_source_id(data);
2044
2045 pasid_present = dma_frcd_pasid_present(data);
2046 guest_addr = dmar_readq(iommu->reg + reg +
2047 fault_index * PRIMARY_FAULT_REG_LEN);
2048 guest_addr = dma_frcd_page_addr(guest_addr);
2049 }
2050
2051 /* clear the fault */
2052 writel(DMA_FRCD_F, iommu->reg + reg +
2053 fault_index * PRIMARY_FAULT_REG_LEN + 12);
2054
2055 raw_spin_unlock_irqrestore(&iommu->register_lock, flag);
2056
2057 if (!ratelimited)
2058 /* Using pasid -1 if pasid is not present */
2059 dmar_fault_do_one(iommu, type, fault_reason,
2060 pasid_present ? pasid : IOMMU_PASID_INVALID,
2061 source_id, guest_addr);
2062
2063 fault_index++;
2064 if (fault_index >= cap_num_fault_regs(iommu->cap))
2065 fault_index = 0;
2066 raw_spin_lock_irqsave(&iommu->register_lock, flag);
2067 }
2068
2069 writel(DMA_FSTS_PFO | DMA_FSTS_PPF | DMA_FSTS_PRO,
2070 iommu->reg + DMAR_FSTS_REG);
2071
2072unlock_exit:
2073 raw_spin_unlock_irqrestore(&iommu->register_lock, flag);
2074 return IRQ_HANDLED;
2075}
2076
2077int dmar_set_interrupt(struct intel_iommu *iommu)
2078{
2079 int irq, ret;
2080
2081 /*
2082 * Check if the fault interrupt is already initialized.
2083 */
2084 if (iommu->irq)
2085 return 0;
2086
2087 irq = dmar_alloc_hwirq(iommu->seq_id, iommu->node, iommu);
2088 if (irq > 0) {
2089 iommu->irq = irq;
2090 } else {
2091 pr_err("No free IRQ vectors\n");
2092 return -EINVAL;
2093 }
2094
2095 ret = request_irq(irq, dmar_fault, IRQF_NO_THREAD, iommu->name, iommu);
2096 if (ret)
2097 pr_err("Can't request irq\n");
2098 return ret;
2099}
2100
2101int __init enable_drhd_fault_handling(void)
2102{
2103 struct dmar_drhd_unit *drhd;
2104 struct intel_iommu *iommu;
2105
2106 /*
2107 * Enable fault control interrupt.
2108 */
2109 for_each_iommu(iommu, drhd) {
2110 u32 fault_status;
2111 int ret = dmar_set_interrupt(iommu);
2112
2113 if (ret) {
2114 pr_err("DRHD %Lx: failed to enable fault, interrupt, ret %d\n",
2115 (unsigned long long)drhd->reg_base_addr, ret);
2116 return -1;
2117 }
2118
2119 /*
2120 * Clear any previous faults.
2121 */
2122 dmar_fault(iommu->irq, iommu);
2123 fault_status = readl(iommu->reg + DMAR_FSTS_REG);
2124 writel(fault_status, iommu->reg + DMAR_FSTS_REG);
2125 }
2126
2127 return 0;
2128}
2129
2130/*
2131 * Re-enable Queued Invalidation interface.
2132 */
2133int dmar_reenable_qi(struct intel_iommu *iommu)
2134{
2135 if (!ecap_qis(iommu->ecap))
2136 return -ENOENT;
2137
2138 if (!iommu->qi)
2139 return -ENOENT;
2140
2141 /*
2142 * First disable queued invalidation.
2143 */
2144 dmar_disable_qi(iommu);
2145 /*
2146 * Then enable queued invalidation again. Since there is no pending
2147 * invalidation requests now, it's safe to re-enable queued
2148 * invalidation.
2149 */
2150 __dmar_enable_qi(iommu);
2151
2152 return 0;
2153}
2154
2155/*
2156 * Check interrupt remapping support in DMAR table description.
2157 */
2158int __init dmar_ir_support(void)
2159{
2160 struct acpi_table_dmar *dmar;
2161 dmar = (struct acpi_table_dmar *)dmar_tbl;
2162 if (!dmar)
2163 return 0;
2164 return dmar->flags & 0x1;
2165}
2166
2167/* Check whether DMAR units are in use */
2168static inline bool dmar_in_use(void)
2169{
2170 return irq_remapping_enabled || intel_iommu_enabled;
2171}
2172
2173static int __init dmar_free_unused_resources(void)
2174{
2175 struct dmar_drhd_unit *dmaru, *dmaru_n;
2176
2177 if (dmar_in_use())
2178 return 0;
2179
2180 if (dmar_dev_scope_status != 1 && !list_empty(&dmar_drhd_units))
2181 bus_unregister_notifier(&pci_bus_type, &dmar_pci_bus_nb);
2182
2183 down_write(&dmar_global_lock);
2184 list_for_each_entry_safe(dmaru, dmaru_n, &dmar_drhd_units, list) {
2185 list_del(&dmaru->list);
2186 dmar_free_drhd(dmaru);
2187 }
2188 up_write(&dmar_global_lock);
2189
2190 return 0;
2191}
2192
2193late_initcall(dmar_free_unused_resources);
2194
2195/*
2196 * DMAR Hotplug Support
2197 * For more details, please refer to Intel(R) Virtualization Technology
2198 * for Directed-IO Architecture Specifiction, Rev 2.2, Section 8.8
2199 * "Remapping Hardware Unit Hot Plug".
2200 */
2201static guid_t dmar_hp_guid =
2202 GUID_INIT(0xD8C1A3A6, 0xBE9B, 0x4C9B,
2203 0x91, 0xBF, 0xC3, 0xCB, 0x81, 0xFC, 0x5D, 0xAF);
2204
2205/*
2206 * Currently there's only one revision and BIOS will not check the revision id,
2207 * so use 0 for safety.
2208 */
2209#define DMAR_DSM_REV_ID 0
2210#define DMAR_DSM_FUNC_DRHD 1
2211#define DMAR_DSM_FUNC_ATSR 2
2212#define DMAR_DSM_FUNC_RHSA 3
2213#define DMAR_DSM_FUNC_SATC 4
2214
2215static inline bool dmar_detect_dsm(acpi_handle handle, int func)
2216{
2217 return acpi_check_dsm(handle, &dmar_hp_guid, DMAR_DSM_REV_ID, 1 << func);
2218}
2219
2220static int dmar_walk_dsm_resource(acpi_handle handle, int func,
2221 dmar_res_handler_t handler, void *arg)
2222{
2223 int ret = -ENODEV;
2224 union acpi_object *obj;
2225 struct acpi_dmar_header *start;
2226 struct dmar_res_callback callback;
2227 static int res_type[] = {
2228 [DMAR_DSM_FUNC_DRHD] = ACPI_DMAR_TYPE_HARDWARE_UNIT,
2229 [DMAR_DSM_FUNC_ATSR] = ACPI_DMAR_TYPE_ROOT_ATS,
2230 [DMAR_DSM_FUNC_RHSA] = ACPI_DMAR_TYPE_HARDWARE_AFFINITY,
2231 [DMAR_DSM_FUNC_SATC] = ACPI_DMAR_TYPE_SATC,
2232 };
2233
2234 if (!dmar_detect_dsm(handle, func))
2235 return 0;
2236
2237 obj = acpi_evaluate_dsm_typed(handle, &dmar_hp_guid, DMAR_DSM_REV_ID,
2238 func, NULL, ACPI_TYPE_BUFFER);
2239 if (!obj)
2240 return -ENODEV;
2241
2242 memset(&callback, 0, sizeof(callback));
2243 callback.cb[res_type[func]] = handler;
2244 callback.arg[res_type[func]] = arg;
2245 start = (struct acpi_dmar_header *)obj->buffer.pointer;
2246 ret = dmar_walk_remapping_entries(start, obj->buffer.length, &callback);
2247
2248 ACPI_FREE(obj);
2249
2250 return ret;
2251}
2252
2253static int dmar_hp_add_drhd(struct acpi_dmar_header *header, void *arg)
2254{
2255 int ret;
2256 struct dmar_drhd_unit *dmaru;
2257
2258 dmaru = dmar_find_dmaru((struct acpi_dmar_hardware_unit *)header);
2259 if (!dmaru)
2260 return -ENODEV;
2261
2262 ret = dmar_ir_hotplug(dmaru, true);
2263 if (ret == 0)
2264 ret = dmar_iommu_hotplug(dmaru, true);
2265
2266 return ret;
2267}
2268
2269static int dmar_hp_remove_drhd(struct acpi_dmar_header *header, void *arg)
2270{
2271 int i, ret;
2272 struct device *dev;
2273 struct dmar_drhd_unit *dmaru;
2274
2275 dmaru = dmar_find_dmaru((struct acpi_dmar_hardware_unit *)header);
2276 if (!dmaru)
2277 return 0;
2278
2279 /*
2280 * All PCI devices managed by this unit should have been destroyed.
2281 */
2282 if (!dmaru->include_all && dmaru->devices && dmaru->devices_cnt) {
2283 for_each_active_dev_scope(dmaru->devices,
2284 dmaru->devices_cnt, i, dev)
2285 return -EBUSY;
2286 }
2287
2288 ret = dmar_ir_hotplug(dmaru, false);
2289 if (ret == 0)
2290 ret = dmar_iommu_hotplug(dmaru, false);
2291
2292 return ret;
2293}
2294
2295static int dmar_hp_release_drhd(struct acpi_dmar_header *header, void *arg)
2296{
2297 struct dmar_drhd_unit *dmaru;
2298
2299 dmaru = dmar_find_dmaru((struct acpi_dmar_hardware_unit *)header);
2300 if (dmaru) {
2301 list_del_rcu(&dmaru->list);
2302 synchronize_rcu();
2303 dmar_free_drhd(dmaru);
2304 }
2305
2306 return 0;
2307}
2308
2309static int dmar_hotplug_insert(acpi_handle handle)
2310{
2311 int ret;
2312 int drhd_count = 0;
2313
2314 ret = dmar_walk_dsm_resource(handle, DMAR_DSM_FUNC_DRHD,
2315 &dmar_validate_one_drhd, (void *)1);
2316 if (ret)
2317 goto out;
2318
2319 ret = dmar_walk_dsm_resource(handle, DMAR_DSM_FUNC_DRHD,
2320 &dmar_parse_one_drhd, (void *)&drhd_count);
2321 if (ret == 0 && drhd_count == 0) {
2322 pr_warn(FW_BUG "No DRHD structures in buffer returned by _DSM method\n");
2323 goto out;
2324 } else if (ret) {
2325 goto release_drhd;
2326 }
2327
2328 ret = dmar_walk_dsm_resource(handle, DMAR_DSM_FUNC_RHSA,
2329 &dmar_parse_one_rhsa, NULL);
2330 if (ret)
2331 goto release_drhd;
2332
2333 ret = dmar_walk_dsm_resource(handle, DMAR_DSM_FUNC_ATSR,
2334 &dmar_parse_one_atsr, NULL);
2335 if (ret)
2336 goto release_atsr;
2337
2338 ret = dmar_walk_dsm_resource(handle, DMAR_DSM_FUNC_DRHD,
2339 &dmar_hp_add_drhd, NULL);
2340 if (!ret)
2341 return 0;
2342
2343 dmar_walk_dsm_resource(handle, DMAR_DSM_FUNC_DRHD,
2344 &dmar_hp_remove_drhd, NULL);
2345release_atsr:
2346 dmar_walk_dsm_resource(handle, DMAR_DSM_FUNC_ATSR,
2347 &dmar_release_one_atsr, NULL);
2348release_drhd:
2349 dmar_walk_dsm_resource(handle, DMAR_DSM_FUNC_DRHD,
2350 &dmar_hp_release_drhd, NULL);
2351out:
2352 return ret;
2353}
2354
2355static int dmar_hotplug_remove(acpi_handle handle)
2356{
2357 int ret;
2358
2359 ret = dmar_walk_dsm_resource(handle, DMAR_DSM_FUNC_ATSR,
2360 &dmar_check_one_atsr, NULL);
2361 if (ret)
2362 return ret;
2363
2364 ret = dmar_walk_dsm_resource(handle, DMAR_DSM_FUNC_DRHD,
2365 &dmar_hp_remove_drhd, NULL);
2366 if (ret == 0) {
2367 WARN_ON(dmar_walk_dsm_resource(handle, DMAR_DSM_FUNC_ATSR,
2368 &dmar_release_one_atsr, NULL));
2369 WARN_ON(dmar_walk_dsm_resource(handle, DMAR_DSM_FUNC_DRHD,
2370 &dmar_hp_release_drhd, NULL));
2371 } else {
2372 dmar_walk_dsm_resource(handle, DMAR_DSM_FUNC_DRHD,
2373 &dmar_hp_add_drhd, NULL);
2374 }
2375
2376 return ret;
2377}
2378
2379static acpi_status dmar_get_dsm_handle(acpi_handle handle, u32 lvl,
2380 void *context, void **retval)
2381{
2382 acpi_handle *phdl = retval;
2383
2384 if (dmar_detect_dsm(handle, DMAR_DSM_FUNC_DRHD)) {
2385 *phdl = handle;
2386 return AE_CTRL_TERMINATE;
2387 }
2388
2389 return AE_OK;
2390}
2391
2392static int dmar_device_hotplug(acpi_handle handle, bool insert)
2393{
2394 int ret;
2395 acpi_handle tmp = NULL;
2396 acpi_status status;
2397
2398 if (!dmar_in_use())
2399 return 0;
2400
2401 if (dmar_detect_dsm(handle, DMAR_DSM_FUNC_DRHD)) {
2402 tmp = handle;
2403 } else {
2404 status = acpi_walk_namespace(ACPI_TYPE_DEVICE, handle,
2405 ACPI_UINT32_MAX,
2406 dmar_get_dsm_handle,
2407 NULL, NULL, &tmp);
2408 if (ACPI_FAILURE(status)) {
2409 pr_warn("Failed to locate _DSM method.\n");
2410 return -ENXIO;
2411 }
2412 }
2413 if (tmp == NULL)
2414 return 0;
2415
2416 down_write(&dmar_global_lock);
2417 if (insert)
2418 ret = dmar_hotplug_insert(tmp);
2419 else
2420 ret = dmar_hotplug_remove(tmp);
2421 up_write(&dmar_global_lock);
2422
2423 return ret;
2424}
2425
2426int dmar_device_add(acpi_handle handle)
2427{
2428 return dmar_device_hotplug(handle, true);
2429}
2430
2431int dmar_device_remove(acpi_handle handle)
2432{
2433 return dmar_device_hotplug(handle, false);
2434}
2435
2436/*
2437 * dmar_platform_optin - Is %DMA_CTRL_PLATFORM_OPT_IN_FLAG set in DMAR table
2438 *
2439 * Returns true if the platform has %DMA_CTRL_PLATFORM_OPT_IN_FLAG set in
2440 * the ACPI DMAR table. This means that the platform boot firmware has made
2441 * sure no device can issue DMA outside of RMRR regions.
2442 */
2443bool dmar_platform_optin(void)
2444{
2445 struct acpi_table_dmar *dmar;
2446 acpi_status status;
2447 bool ret;
2448
2449 status = acpi_get_table(ACPI_SIG_DMAR, 0,
2450 (struct acpi_table_header **)&dmar);
2451 if (ACPI_FAILURE(status))
2452 return false;
2453
2454 ret = !!(dmar->flags & DMAR_PLATFORM_OPT_IN);
2455 acpi_put_table((struct acpi_table_header *)dmar);
2456
2457 return ret;
2458}
2459EXPORT_SYMBOL_GPL(dmar_platform_optin);