summaryrefslogtreecommitdiff
path: root/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c
blob: 20436f81856ad280f112bf52dd42ea6157443b04 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
/*
 * Copyright 2019 Advanced Micro Devices, Inc.
 *
 * Permission is hereby granted, free of charge, to any person obtaining a
 * copy of this software and associated documentation files (the "Software"),
 * to deal in the Software without restriction, including without limitation
 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
 * and/or sell copies of the Software, and to permit persons to whom the
 * Software is furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice shall be included in
 * all copies or substantial portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
 * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
 * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
 * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
 * OTHER DEALINGS IN THE SOFTWARE.
 *
 */

#include "amdgpu.h"
#include "umc_v6_7.h"
#define MAX_UMC_POISON_POLLING_TIME_SYNC   20  //ms

static int amdgpu_umc_convert_error_address(struct amdgpu_device *adev,
				    struct ras_err_data *err_data, uint64_t err_addr,
				    uint32_t ch_inst, uint32_t umc_inst)
{
	switch (amdgpu_ip_version(adev, UMC_HWIP, 0)) {
	case IP_VERSION(6, 7, 0):
		umc_v6_7_convert_error_address(adev,
				err_data, err_addr, ch_inst, umc_inst);
		break;
	default:
		dev_warn(adev->dev,
			 "UMC address to Physical address translation is not supported\n");
		return AMDGPU_RAS_FAIL;
	}

	return AMDGPU_RAS_SUCCESS;
}

int amdgpu_umc_page_retirement_mca(struct amdgpu_device *adev,
			uint64_t err_addr, uint32_t ch_inst, uint32_t umc_inst)
{
	struct ras_err_data err_data;
	int ret;

	ret = amdgpu_ras_error_data_init(&err_data);
	if (ret)
		return ret;

	err_data.err_addr =
		kcalloc(adev->umc.max_ras_err_cnt_per_query,
			sizeof(struct eeprom_table_record), GFP_KERNEL);
	if (!err_data.err_addr) {
		dev_warn(adev->dev,
			"Failed to alloc memory for umc error record in MCA notifier!\n");
		ret = AMDGPU_RAS_FAIL;
		goto out_fini_err_data;
	}

	/*
	 * Translate UMC channel address to Physical address
	 */
	ret = amdgpu_umc_convert_error_address(adev, &err_data, err_addr,
					ch_inst, umc_inst);
	if (ret)
		goto out_free_err_addr;

	if (amdgpu_bad_page_threshold != 0) {
		amdgpu_ras_add_bad_pages(adev, err_data.err_addr,
						err_data.err_addr_cnt);
		amdgpu_ras_save_bad_pages(adev, NULL);
	}

out_free_err_addr:
	kfree(err_data.err_addr);

out_fini_err_data:
	amdgpu_ras_error_data_fini(&err_data);

	return ret;
}

static void amdgpu_umc_handle_bad_pages(struct amdgpu_device *adev,
			void *ras_error_status)
{
	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
	unsigned int error_query_mode;
	int ret = 0;
	unsigned long err_count;

	amdgpu_ras_get_error_query_mode(adev, &error_query_mode);

	mutex_lock(&con->page_retirement_lock);
	ret = amdgpu_dpm_get_ecc_info(adev, (void *)&(con->umc_ecc));
	if (ret == -EOPNOTSUPP &&
	    error_query_mode == AMDGPU_RAS_DIRECT_ERROR_QUERY) {
		if (adev->umc.ras && adev->umc.ras->ras_block.hw_ops &&
		    adev->umc.ras->ras_block.hw_ops->query_ras_error_count)
		    adev->umc.ras->ras_block.hw_ops->query_ras_error_count(adev, ras_error_status);

		if (adev->umc.ras && adev->umc.ras->ras_block.hw_ops &&
		    adev->umc.ras->ras_block.hw_ops->query_ras_error_address &&
		    adev->umc.max_ras_err_cnt_per_query) {
			err_data->err_addr =
				kcalloc(adev->umc.max_ras_err_cnt_per_query,
					sizeof(struct eeprom_table_record), GFP_KERNEL);

			/* still call query_ras_error_address to clear error status
			 * even NOMEM error is encountered
			 */
			if(!err_data->err_addr)
				dev_warn(adev->dev, "Failed to alloc memory for "
						"umc error address record!\n");

			/* umc query_ras_error_address is also responsible for clearing
			 * error status
			 */
			adev->umc.ras->ras_block.hw_ops->query_ras_error_address(adev, ras_error_status);
		}
	} else if (error_query_mode == AMDGPU_RAS_FIRMWARE_ERROR_QUERY ||
	    (!ret && error_query_mode == AMDGPU_RAS_DIRECT_ERROR_QUERY)) {
		if (adev->umc.ras &&
		    adev->umc.ras->ecc_info_query_ras_error_count)
		    adev->umc.ras->ecc_info_query_ras_error_count(adev, ras_error_status);

		if (adev->umc.ras &&
		    adev->umc.ras->ecc_info_query_ras_error_address &&
		    adev->umc.max_ras_err_cnt_per_query) {
			err_data->err_addr =
				kcalloc(adev->umc.max_ras_err_cnt_per_query,
					sizeof(struct eeprom_table_record), GFP_KERNEL);

			/* still call query_ras_error_address to clear error status
			 * even NOMEM error is encountered
			 */
			if(!err_data->err_addr)
				dev_warn(adev->dev, "Failed to alloc memory for "
						"umc error address record!\n");

			/* umc query_ras_error_address is also responsible for clearing
			 * error status
			 */
			adev->umc.ras->ecc_info_query_ras_error_address(adev, ras_error_status);
		}
	}

	/* only uncorrectable error needs gpu reset */
	if (err_data->ue_count || err_data->de_count) {
		err_count = err_data->ue_count + err_data->de_count;
		if ((amdgpu_bad_page_threshold != 0) &&
			err_data->err_addr_cnt) {
			amdgpu_ras_add_bad_pages(adev, err_data->err_addr,
						err_data->err_addr_cnt);
			amdgpu_ras_save_bad_pages(adev, &err_count);

			amdgpu_dpm_send_hbm_bad_pages_num(adev, con->eeprom_control.ras_num_recs);

			if (con->update_channel_flag == true) {
				amdgpu_dpm_send_hbm_bad_channel_flag(adev, con->eeprom_control.bad_channel_bitmap);
				con->update_channel_flag = false;
			}
		}
	}

	kfree(err_data->err_addr);

	mutex_unlock(&con->page_retirement_lock);
}

static int amdgpu_umc_do_page_retirement(struct amdgpu_device *adev,
		void *ras_error_status,
		struct amdgpu_iv_entry *entry,
		bool reset)
{
	struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status;
	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);

	kgd2kfd_set_sram_ecc_flag(adev->kfd.dev);
	amdgpu_umc_handle_bad_pages(adev, ras_error_status);

	if (err_data->ue_count && reset) {
		/* use mode-2 reset for poison consumption */
		if (!entry)
			con->gpu_reset_flags |= AMDGPU_RAS_GPU_RESET_MODE2_RESET;
		amdgpu_ras_reset_gpu(adev);
	}

	return AMDGPU_RAS_SUCCESS;
}

int amdgpu_umc_bad_page_polling_timeout(struct amdgpu_device *adev,
			bool reset, uint32_t timeout_ms)
{
	struct ras_err_data err_data;
	struct ras_common_if head = {
		.block = AMDGPU_RAS_BLOCK__UMC,
	};
	struct ras_manager *obj = amdgpu_ras_find_obj(adev, &head);
	uint32_t timeout = timeout_ms;

	memset(&err_data, 0, sizeof(err_data));
	amdgpu_ras_error_data_init(&err_data);

	do {

		amdgpu_umc_handle_bad_pages(adev, &err_data);

		if (timeout && !err_data.de_count) {
			msleep(1);
			timeout--;
		}

	} while (timeout && !err_data.de_count);

	if (!timeout)
		dev_warn(adev->dev, "Can't find bad pages\n");

	if (err_data.de_count)
		dev_info(adev->dev, "%ld new deferred hardware errors detected\n", err_data.de_count);

	if (obj) {
		obj->err_data.ue_count += err_data.ue_count;
		obj->err_data.ce_count += err_data.ce_count;
		obj->err_data.de_count += err_data.de_count;
	}

	amdgpu_ras_error_data_fini(&err_data);

	kgd2kfd_set_sram_ecc_flag(adev->kfd.dev);

	if (reset) {
		struct amdgpu_ras *con = amdgpu_ras_get_context(adev);

		/* use mode-2 reset for poison consumption */
		con->gpu_reset_flags |= AMDGPU_RAS_GPU_RESET_MODE2_RESET;
		amdgpu_ras_reset_gpu(adev);
	}

	return 0;
}

int amdgpu_umc_poison_handler(struct amdgpu_device *adev,
			enum amdgpu_ras_block block, bool reset)
{
	int ret = AMDGPU_RAS_SUCCESS;

	if (adev->gmc.xgmi.connected_to_cpu ||
		adev->gmc.is_app_apu) {
		if (reset) {
			/* MCA poison handler is only responsible for GPU reset,
			 * let MCA notifier do page retirement.
			 */
			kgd2kfd_set_sram_ecc_flag(adev->kfd.dev);
			amdgpu_ras_reset_gpu(adev);
		}
		return ret;
	}

	if (!amdgpu_sriov_vf(adev)) {
		if (amdgpu_ip_version(adev, UMC_HWIP, 0) < IP_VERSION(12, 0, 0)) {
			struct ras_err_data err_data;
			struct ras_common_if head = {
				.block = AMDGPU_RAS_BLOCK__UMC,
			};
			struct ras_manager *obj = amdgpu_ras_find_obj(adev, &head);

			ret = amdgpu_ras_error_data_init(&err_data);
			if (ret)
				return ret;

			ret = amdgpu_umc_do_page_retirement(adev, &err_data, NULL, reset);

			if (ret == AMDGPU_RAS_SUCCESS && obj) {
				obj->err_data.ue_count += err_data.ue_count;
				obj->err_data.ce_count += err_data.ce_count;
				obj->err_data.de_count += err_data.de_count;
			}

			amdgpu_ras_error_data_fini(&err_data);
		} else {
			if (reset) {
				amdgpu_umc_bad_page_polling_timeout(adev,
							reset, MAX_UMC_POISON_POLLING_TIME_SYNC);
			} else {
				struct amdgpu_ras *con = amdgpu_ras_get_context(adev);

				atomic_inc(&con->page_retirement_req_cnt);

				wake_up(&con->page_retirement_wq);
			}
		}
	} else {
		if (adev->virt.ops && adev->virt.ops->ras_poison_handler)
			adev->virt.ops->ras_poison_handler(adev, block);
		else
			dev_warn(adev->dev,
				"No ras_poison_handler interface in SRIOV!\n");
	}

	return ret;
}

int amdgpu_umc_process_ras_data_cb(struct amdgpu_device *adev,
		void *ras_error_status,
		struct amdgpu_iv_entry *entry)
{
	return amdgpu_umc_do_page_retirement(adev, ras_error_status, entry, true);
}

int amdgpu_umc_ras_sw_init(struct amdgpu_device *adev)
{
	int err;
	struct amdgpu_umc_ras *ras;

	if (!adev->umc.ras)
		return 0;

	ras = adev->umc.ras;

	err = amdgpu_ras_register_ras_block(adev, &ras->ras_block);
	if (err) {
		dev_err(adev->dev, "Failed to register umc ras block!\n");
		return err;
	}

	strcpy(adev->umc.ras->ras_block.ras_comm.name, "umc");
	ras->ras_block.ras_comm.block = AMDGPU_RAS_BLOCK__UMC;
	ras->ras_block.ras_comm.type = AMDGPU_RAS_ERROR__MULTI_UNCORRECTABLE;
	adev->umc.ras_if = &ras->ras_block.ras_comm;

	if (!ras->ras_block.ras_late_init)
		ras->ras_block.ras_late_init = amdgpu_umc_ras_late_init;

	if (!ras->ras_block.ras_cb)
		ras->ras_block.ras_cb = amdgpu_umc_process_ras_data_cb;

	return 0;
}

int amdgpu_umc_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block)
{
	int r;

	r = amdgpu_ras_block_late_init(adev, ras_block);
	if (r)
		return r;

	if (amdgpu_ras_is_supported(adev, ras_block->block)) {
		r = amdgpu_irq_get(adev, &adev->gmc.ecc_irq, 0);
		if (r)
			goto late_fini;
	}

	/* ras init of specific umc version */
	if (adev->umc.ras &&
	    adev->umc.ras->err_cnt_init)
		adev->umc.ras->err_cnt_init(adev);

	return 0;

late_fini:
	amdgpu_ras_block_late_fini(adev, ras_block);
	return r;
}

int amdgpu_umc_process_ecc_irq(struct amdgpu_device *adev,
		struct amdgpu_irq_src *source,
		struct amdgpu_iv_entry *entry)
{
	struct ras_common_if *ras_if = adev->umc.ras_if;
	struct ras_dispatch_if ih_data = {
		.entry = entry,
	};

	if (!ras_if)
		return 0;

	ih_data.head = *ras_if;

	amdgpu_ras_interrupt_dispatch(adev, &ih_data);
	return 0;
}

void amdgpu_umc_fill_error_record(struct ras_err_data *err_data,
		uint64_t err_addr,
		uint64_t retired_page,
		uint32_t channel_index,
		uint32_t umc_inst)
{
	struct eeprom_table_record *err_rec =
		&err_data->err_addr[err_data->err_addr_cnt];

	err_rec->address = err_addr;
	/* page frame address is saved */
	err_rec->retired_page = retired_page >> AMDGPU_GPU_PAGE_SHIFT;
	err_rec->ts = (uint64_t)ktime_get_real_seconds();
	err_rec->err_type = AMDGPU_RAS_EEPROM_ERR_NON_RECOVERABLE;
	err_rec->cu = 0;
	err_rec->mem_channel = channel_index;
	err_rec->mcumc_id = umc_inst;

	err_data->err_addr_cnt++;
}

int amdgpu_umc_loop_channels(struct amdgpu_device *adev,
			umc_func func, void *data)
{
	uint32_t node_inst       = 0;
	uint32_t umc_inst        = 0;
	uint32_t ch_inst         = 0;
	int ret = 0;

	if (adev->umc.node_inst_num) {
		LOOP_UMC_EACH_NODE_INST_AND_CH(node_inst, umc_inst, ch_inst) {
			ret = func(adev, node_inst, umc_inst, ch_inst, data);
			if (ret) {
				dev_err(adev->dev, "Node %d umc %d ch %d func returns %d\n",
					node_inst, umc_inst, ch_inst, ret);
				return ret;
			}
		}
	} else {
		LOOP_UMC_INST_AND_CH(umc_inst, ch_inst) {
			ret = func(adev, 0, umc_inst, ch_inst, data);
			if (ret) {
				dev_err(adev->dev, "Umc %d ch %d func returns %d\n",
					umc_inst, ch_inst, ret);
				return ret;
			}
		}
	}

	return 0;
}