Loading...
1/*
2 * APEI Hardware Error Souce Table support
3 *
4 * HEST describes error sources in detail; communicates operational
5 * parameters (i.e. severity levels, masking bits, and threshold
6 * values) to Linux as necessary. It also allows the BIOS to report
7 * non-standard error sources to Linux (for example, chipset-specific
8 * error registers).
9 *
10 * For more information about HEST, please refer to ACPI Specification
11 * version 4.0, section 17.3.2.
12 *
13 * Copyright 2009 Intel Corp.
14 * Author: Huang Ying <ying.huang@intel.com>
15 *
16 * This program is free software; you can redistribute it and/or
17 * modify it under the terms of the GNU General Public License version
18 * 2 as published by the Free Software Foundation;
19 *
20 * This program is distributed in the hope that it will be useful,
21 * but WITHOUT ANY WARRANTY; without even the implied warranty of
22 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
23 * GNU General Public License for more details.
24 *
25 * You should have received a copy of the GNU General Public License
26 * along with this program; if not, write to the Free Software
27 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
28 */
29
30#include <linux/kernel.h>
31#include <linux/module.h>
32#include <linux/init.h>
33#include <linux/acpi.h>
34#include <linux/kdebug.h>
35#include <linux/highmem.h>
36#include <linux/io.h>
37#include <linux/platform_device.h>
38#include <acpi/apei.h>
39
40#include "apei-internal.h"
41
42#define HEST_PFX "HEST: "
43
44bool hest_disable;
45EXPORT_SYMBOL_GPL(hest_disable);
46
47/* HEST table parsing */
48
49static struct acpi_table_hest *__read_mostly hest_tab;
50
51static const int hest_esrc_len_tab[ACPI_HEST_TYPE_RESERVED] = {
52 [ACPI_HEST_TYPE_IA32_CHECK] = -1, /* need further calculation */
53 [ACPI_HEST_TYPE_IA32_CORRECTED_CHECK] = -1,
54 [ACPI_HEST_TYPE_IA32_NMI] = sizeof(struct acpi_hest_ia_nmi),
55 [ACPI_HEST_TYPE_AER_ROOT_PORT] = sizeof(struct acpi_hest_aer_root),
56 [ACPI_HEST_TYPE_AER_ENDPOINT] = sizeof(struct acpi_hest_aer),
57 [ACPI_HEST_TYPE_AER_BRIDGE] = sizeof(struct acpi_hest_aer_bridge),
58 [ACPI_HEST_TYPE_GENERIC_ERROR] = sizeof(struct acpi_hest_generic),
59};
60
61static int hest_esrc_len(struct acpi_hest_header *hest_hdr)
62{
63 u16 hest_type = hest_hdr->type;
64 int len;
65
66 if (hest_type >= ACPI_HEST_TYPE_RESERVED)
67 return 0;
68
69 len = hest_esrc_len_tab[hest_type];
70
71 if (hest_type == ACPI_HEST_TYPE_IA32_CORRECTED_CHECK) {
72 struct acpi_hest_ia_corrected *cmc;
73 cmc = (struct acpi_hest_ia_corrected *)hest_hdr;
74 len = sizeof(*cmc) + cmc->num_hardware_banks *
75 sizeof(struct acpi_hest_ia_error_bank);
76 } else if (hest_type == ACPI_HEST_TYPE_IA32_CHECK) {
77 struct acpi_hest_ia_machine_check *mc;
78 mc = (struct acpi_hest_ia_machine_check *)hest_hdr;
79 len = sizeof(*mc) + mc->num_hardware_banks *
80 sizeof(struct acpi_hest_ia_error_bank);
81 }
82 BUG_ON(len == -1);
83
84 return len;
85};
86
87int apei_hest_parse(apei_hest_func_t func, void *data)
88{
89 struct acpi_hest_header *hest_hdr;
90 int i, rc, len;
91
92 if (hest_disable)
93 return -EINVAL;
94
95 hest_hdr = (struct acpi_hest_header *)(hest_tab + 1);
96 for (i = 0; i < hest_tab->error_source_count; i++) {
97 len = hest_esrc_len(hest_hdr);
98 if (!len) {
99 pr_warning(FW_WARN HEST_PFX
100 "Unknown or unused hardware error source "
101 "type: %d for hardware error source: %d.\n",
102 hest_hdr->type, hest_hdr->source_id);
103 return -EINVAL;
104 }
105 if ((void *)hest_hdr + len >
106 (void *)hest_tab + hest_tab->header.length) {
107 pr_warning(FW_BUG HEST_PFX
108 "Table contents overflow for hardware error source: %d.\n",
109 hest_hdr->source_id);
110 return -EINVAL;
111 }
112
113 rc = func(hest_hdr, data);
114 if (rc)
115 return rc;
116
117 hest_hdr = (void *)hest_hdr + len;
118 }
119
120 return 0;
121}
122EXPORT_SYMBOL_GPL(apei_hest_parse);
123
124struct ghes_arr {
125 struct platform_device **ghes_devs;
126 unsigned int count;
127};
128
129static int __init hest_parse_ghes_count(struct acpi_hest_header *hest_hdr, void *data)
130{
131 int *count = data;
132
133 if (hest_hdr->type == ACPI_HEST_TYPE_GENERIC_ERROR)
134 (*count)++;
135 return 0;
136}
137
138static int __init hest_parse_ghes(struct acpi_hest_header *hest_hdr, void *data)
139{
140 struct platform_device *ghes_dev;
141 struct ghes_arr *ghes_arr = data;
142 int rc, i;
143
144 if (hest_hdr->type != ACPI_HEST_TYPE_GENERIC_ERROR)
145 return 0;
146
147 if (!((struct acpi_hest_generic *)hest_hdr)->enabled)
148 return 0;
149 for (i = 0; i < ghes_arr->count; i++) {
150 struct acpi_hest_header *hdr;
151 ghes_dev = ghes_arr->ghes_devs[i];
152 hdr = *(struct acpi_hest_header **)ghes_dev->dev.platform_data;
153 if (hdr->source_id == hest_hdr->source_id) {
154 pr_warning(FW_WARN HEST_PFX "Duplicated hardware error source ID: %d.\n",
155 hdr->source_id);
156 return -EIO;
157 }
158 }
159 ghes_dev = platform_device_alloc("GHES", hest_hdr->source_id);
160 if (!ghes_dev)
161 return -ENOMEM;
162
163 rc = platform_device_add_data(ghes_dev, &hest_hdr, sizeof(void *));
164 if (rc)
165 goto err;
166
167 rc = platform_device_add(ghes_dev);
168 if (rc)
169 goto err;
170 ghes_arr->ghes_devs[ghes_arr->count++] = ghes_dev;
171
172 return 0;
173err:
174 platform_device_put(ghes_dev);
175 return rc;
176}
177
178static int __init hest_ghes_dev_register(unsigned int ghes_count)
179{
180 int rc, i;
181 struct ghes_arr ghes_arr;
182
183 ghes_arr.count = 0;
184 ghes_arr.ghes_devs = kmalloc(sizeof(void *) * ghes_count, GFP_KERNEL);
185 if (!ghes_arr.ghes_devs)
186 return -ENOMEM;
187
188 rc = apei_hest_parse(hest_parse_ghes, &ghes_arr);
189 if (rc)
190 goto err;
191out:
192 kfree(ghes_arr.ghes_devs);
193 return rc;
194err:
195 for (i = 0; i < ghes_arr.count; i++)
196 platform_device_unregister(ghes_arr.ghes_devs[i]);
197 goto out;
198}
199
200static int __init setup_hest_disable(char *str)
201{
202 hest_disable = 1;
203 return 0;
204}
205
206__setup("hest_disable", setup_hest_disable);
207
208void __init acpi_hest_init(void)
209{
210 acpi_status status;
211 int rc = -ENODEV;
212 unsigned int ghes_count = 0;
213
214 if (hest_disable) {
215 pr_info(HEST_PFX "Table parsing disabled.\n");
216 return;
217 }
218
219 if (acpi_disabled)
220 goto err;
221
222 status = acpi_get_table(ACPI_SIG_HEST, 0,
223 (struct acpi_table_header **)&hest_tab);
224 if (status == AE_NOT_FOUND)
225 goto err;
226 else if (ACPI_FAILURE(status)) {
227 const char *msg = acpi_format_exception(status);
228 pr_err(HEST_PFX "Failed to get table, %s\n", msg);
229 rc = -EINVAL;
230 goto err;
231 }
232
233 if (!ghes_disable) {
234 rc = apei_hest_parse(hest_parse_ghes_count, &ghes_count);
235 if (rc)
236 goto err;
237 rc = hest_ghes_dev_register(ghes_count);
238 if (rc)
239 goto err;
240 }
241
242 pr_info(HEST_PFX "Table parsing has been initialized.\n");
243 return;
244err:
245 hest_disable = 1;
246}
1// SPDX-License-Identifier: GPL-2.0-only
2/*
3 * APEI Hardware Error Source Table support
4 *
5 * HEST describes error sources in detail; communicates operational
6 * parameters (i.e. severity levels, masking bits, and threshold
7 * values) to Linux as necessary. It also allows the BIOS to report
8 * non-standard error sources to Linux (for example, chipset-specific
9 * error registers).
10 *
11 * For more information about HEST, please refer to ACPI Specification
12 * version 4.0, section 17.3.2.
13 *
14 * Copyright 2009 Intel Corp.
15 * Author: Huang Ying <ying.huang@intel.com>
16 */
17
18#include <linux/kernel.h>
19#include <linux/module.h>
20#include <linux/init.h>
21#include <linux/acpi.h>
22#include <linux/kdebug.h>
23#include <linux/highmem.h>
24#include <linux/io.h>
25#include <linux/platform_device.h>
26#include <acpi/apei.h>
27#include <acpi/ghes.h>
28
29#include "apei-internal.h"
30
31#define HEST_PFX "HEST: "
32
33int hest_disable;
34EXPORT_SYMBOL_GPL(hest_disable);
35
36/* HEST table parsing */
37
38static struct acpi_table_hest *__read_mostly hest_tab;
39
40/*
41 * Since GHES_ASSIST is not supported, skip initialization of GHES_ASSIST
42 * structures for MCA.
43 * During HEST parsing, detected MCA error sources are cached from early
44 * table entries so that the Flags and Source Id fields from these cached
45 * values are then referred to in later table entries to determine if the
46 * encountered GHES_ASSIST structure should be initialized.
47 */
48static struct {
49 struct acpi_hest_ia_corrected *cmc;
50 struct acpi_hest_ia_machine_check *mc;
51 struct acpi_hest_ia_deferred_check *dmc;
52} mces;
53
54static const int hest_esrc_len_tab[ACPI_HEST_TYPE_RESERVED] = {
55 [ACPI_HEST_TYPE_IA32_CHECK] = -1, /* need further calculation */
56 [ACPI_HEST_TYPE_IA32_CORRECTED_CHECK] = -1,
57 [ACPI_HEST_TYPE_IA32_NMI] = sizeof(struct acpi_hest_ia_nmi),
58 [ACPI_HEST_TYPE_AER_ROOT_PORT] = sizeof(struct acpi_hest_aer_root),
59 [ACPI_HEST_TYPE_AER_ENDPOINT] = sizeof(struct acpi_hest_aer),
60 [ACPI_HEST_TYPE_AER_BRIDGE] = sizeof(struct acpi_hest_aer_bridge),
61 [ACPI_HEST_TYPE_GENERIC_ERROR] = sizeof(struct acpi_hest_generic),
62 [ACPI_HEST_TYPE_GENERIC_ERROR_V2] = sizeof(struct acpi_hest_generic_v2),
63 [ACPI_HEST_TYPE_IA32_DEFERRED_CHECK] = -1,
64};
65
66static inline bool is_generic_error(struct acpi_hest_header *hest_hdr)
67{
68 return hest_hdr->type == ACPI_HEST_TYPE_GENERIC_ERROR ||
69 hest_hdr->type == ACPI_HEST_TYPE_GENERIC_ERROR_V2;
70}
71
72static int hest_esrc_len(struct acpi_hest_header *hest_hdr)
73{
74 u16 hest_type = hest_hdr->type;
75 int len;
76
77 if (hest_type >= ACPI_HEST_TYPE_RESERVED)
78 return 0;
79
80 len = hest_esrc_len_tab[hest_type];
81
82 if (hest_type == ACPI_HEST_TYPE_IA32_CORRECTED_CHECK) {
83 struct acpi_hest_ia_corrected *cmc;
84 cmc = (struct acpi_hest_ia_corrected *)hest_hdr;
85 len = sizeof(*cmc) + cmc->num_hardware_banks *
86 sizeof(struct acpi_hest_ia_error_bank);
87 mces.cmc = cmc;
88 } else if (hest_type == ACPI_HEST_TYPE_IA32_CHECK) {
89 struct acpi_hest_ia_machine_check *mc;
90 mc = (struct acpi_hest_ia_machine_check *)hest_hdr;
91 len = sizeof(*mc) + mc->num_hardware_banks *
92 sizeof(struct acpi_hest_ia_error_bank);
93 mces.mc = mc;
94 } else if (hest_type == ACPI_HEST_TYPE_IA32_DEFERRED_CHECK) {
95 struct acpi_hest_ia_deferred_check *mc;
96 mc = (struct acpi_hest_ia_deferred_check *)hest_hdr;
97 len = sizeof(*mc) + mc->num_hardware_banks *
98 sizeof(struct acpi_hest_ia_error_bank);
99 mces.dmc = mc;
100 }
101 BUG_ON(len == -1);
102
103 return len;
104};
105
106/*
107 * GHES and GHESv2 structures share the same format, starting from
108 * Source Id and ending in Error Status Block Length (inclusive).
109 */
110static bool is_ghes_assist_struct(struct acpi_hest_header *hest_hdr)
111{
112 struct acpi_hest_generic *ghes;
113 u16 related_source_id;
114
115 if (hest_hdr->type != ACPI_HEST_TYPE_GENERIC_ERROR &&
116 hest_hdr->type != ACPI_HEST_TYPE_GENERIC_ERROR_V2)
117 return false;
118
119 ghes = (struct acpi_hest_generic *)hest_hdr;
120 related_source_id = ghes->related_source_id;
121
122 if (mces.cmc && mces.cmc->flags & ACPI_HEST_GHES_ASSIST &&
123 related_source_id == mces.cmc->header.source_id)
124 return true;
125 if (mces.mc && mces.mc->flags & ACPI_HEST_GHES_ASSIST &&
126 related_source_id == mces.mc->header.source_id)
127 return true;
128 if (mces.dmc && mces.dmc->flags & ACPI_HEST_GHES_ASSIST &&
129 related_source_id == mces.dmc->header.source_id)
130 return true;
131
132 return false;
133}
134
135typedef int (*apei_hest_func_t)(struct acpi_hest_header *hest_hdr, void *data);
136
137static int apei_hest_parse(apei_hest_func_t func, void *data)
138{
139 struct acpi_hest_header *hest_hdr;
140 int i, rc, len;
141
142 if (hest_disable || !hest_tab)
143 return -EINVAL;
144
145 hest_hdr = (struct acpi_hest_header *)(hest_tab + 1);
146 for (i = 0; i < hest_tab->error_source_count; i++) {
147 len = hest_esrc_len(hest_hdr);
148 if (!len) {
149 pr_warn(FW_WARN HEST_PFX
150 "Unknown or unused hardware error source "
151 "type: %d for hardware error source: %d.\n",
152 hest_hdr->type, hest_hdr->source_id);
153 return -EINVAL;
154 }
155 if ((void *)hest_hdr + len >
156 (void *)hest_tab + hest_tab->header.length) {
157 pr_warn(FW_BUG HEST_PFX
158 "Table contents overflow for hardware error source: %d.\n",
159 hest_hdr->source_id);
160 return -EINVAL;
161 }
162
163 if (is_ghes_assist_struct(hest_hdr)) {
164 hest_hdr = (void *)hest_hdr + len;
165 continue;
166 }
167
168 rc = func(hest_hdr, data);
169 if (rc)
170 return rc;
171
172 hest_hdr = (void *)hest_hdr + len;
173 }
174
175 return 0;
176}
177
178/*
179 * Check if firmware advertises firmware first mode. We need FF bit to be set
180 * along with a set of MC banks which work in FF mode.
181 */
182static int __init hest_parse_cmc(struct acpi_hest_header *hest_hdr, void *data)
183{
184 if (hest_hdr->type != ACPI_HEST_TYPE_IA32_CORRECTED_CHECK)
185 return 0;
186
187 if (!acpi_disable_cmcff)
188 return !arch_apei_enable_cmcff(hest_hdr, data);
189
190 return 0;
191}
192
193struct ghes_arr {
194 struct platform_device **ghes_devs;
195 unsigned int count;
196};
197
198static int __init hest_parse_ghes_count(struct acpi_hest_header *hest_hdr, void *data)
199{
200 int *count = data;
201
202 if (is_generic_error(hest_hdr))
203 (*count)++;
204 return 0;
205}
206
207static int __init hest_parse_ghes(struct acpi_hest_header *hest_hdr, void *data)
208{
209 struct platform_device *ghes_dev;
210 struct ghes_arr *ghes_arr = data;
211 int rc, i;
212
213 if (!is_generic_error(hest_hdr))
214 return 0;
215
216 if (!((struct acpi_hest_generic *)hest_hdr)->enabled)
217 return 0;
218 for (i = 0; i < ghes_arr->count; i++) {
219 struct acpi_hest_header *hdr;
220 ghes_dev = ghes_arr->ghes_devs[i];
221 hdr = *(struct acpi_hest_header **)ghes_dev->dev.platform_data;
222 if (hdr->source_id == hest_hdr->source_id) {
223 pr_warn(FW_WARN HEST_PFX "Duplicated hardware error source ID: %d.\n",
224 hdr->source_id);
225 return -EIO;
226 }
227 }
228 ghes_dev = platform_device_alloc("GHES", hest_hdr->source_id);
229 if (!ghes_dev)
230 return -ENOMEM;
231
232 rc = platform_device_add_data(ghes_dev, &hest_hdr, sizeof(void *));
233 if (rc)
234 goto err;
235
236 rc = platform_device_add(ghes_dev);
237 if (rc)
238 goto err;
239 ghes_arr->ghes_devs[ghes_arr->count++] = ghes_dev;
240
241 return 0;
242err:
243 platform_device_put(ghes_dev);
244 return rc;
245}
246
247static int __init hest_ghes_dev_register(unsigned int ghes_count)
248{
249 int rc, i;
250 struct ghes_arr ghes_arr;
251
252 ghes_arr.count = 0;
253 ghes_arr.ghes_devs = kmalloc_array(ghes_count, sizeof(void *),
254 GFP_KERNEL);
255 if (!ghes_arr.ghes_devs)
256 return -ENOMEM;
257
258 rc = apei_hest_parse(hest_parse_ghes, &ghes_arr);
259 if (rc)
260 goto err;
261
262 rc = ghes_estatus_pool_init(ghes_count);
263 if (rc)
264 goto err;
265
266out:
267 kfree(ghes_arr.ghes_devs);
268 return rc;
269err:
270 for (i = 0; i < ghes_arr.count; i++)
271 platform_device_unregister(ghes_arr.ghes_devs[i]);
272 goto out;
273}
274
275static int __init setup_hest_disable(char *str)
276{
277 hest_disable = HEST_DISABLED;
278 return 1;
279}
280
281__setup("hest_disable", setup_hest_disable);
282
283void __init acpi_hest_init(void)
284{
285 acpi_status status;
286 int rc;
287 unsigned int ghes_count = 0;
288
289 if (hest_disable) {
290 pr_info(HEST_PFX "Table parsing disabled.\n");
291 return;
292 }
293
294 status = acpi_get_table(ACPI_SIG_HEST, 0,
295 (struct acpi_table_header **)&hest_tab);
296 if (status == AE_NOT_FOUND) {
297 hest_disable = HEST_NOT_FOUND;
298 return;
299 } else if (ACPI_FAILURE(status)) {
300 const char *msg = acpi_format_exception(status);
301 pr_err(HEST_PFX "Failed to get table, %s\n", msg);
302 hest_disable = HEST_DISABLED;
303 return;
304 }
305
306 rc = apei_hest_parse(hest_parse_cmc, NULL);
307 if (rc)
308 goto err;
309
310 if (!ghes_disable) {
311 rc = apei_hest_parse(hest_parse_ghes_count, &ghes_count);
312 if (rc)
313 goto err;
314
315 if (ghes_count)
316 rc = hest_ghes_dev_register(ghes_count);
317 if (rc)
318 goto err;
319 }
320
321 pr_info(HEST_PFX "Table parsing has been initialized.\n");
322 return;
323err:
324 hest_disable = HEST_DISABLED;
325 acpi_put_table((struct acpi_table_header *)hest_tab);
326}