xref: /netbsd-src/sbin/nvmectl/logpage.c (revision e66f4605a1ac8e93e0da0c410a7beb388f0109bf)
1 /*	$NetBSD: logpage.c,v 1.11 2023/02/02 08:21:32 mlelstv Exp $	*/
2 
3 /*-
4  * SPDX-License-Identifier: BSD-2-Clause-FreeBSD
5  *
6  * Copyright (c) 2013 EMC Corp.
7  * All rights reserved.
8  *
9  * Copyright (C) 2012-2013 Intel Corporation
10  * All rights reserved.
11  *
12  * Redistribution and use in source and binary forms, with or without
13  * modification, are permitted provided that the following conditions
14  * are met:
15  * 1. Redistributions of source code must retain the above copyright
16  *    notice, this list of conditions and the following disclaimer.
17  * 2. Redistributions in binary form must reproduce the above copyright
18  *    notice, this list of conditions and the following disclaimer in the
19  *    documentation and/or other materials provided with the distribution.
20  *
21  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
22  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
25  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31  * SUCH DAMAGE.
32  */
33 
34 #include <sys/cdefs.h>
35 #ifndef lint
36 __RCSID("$NetBSD: logpage.c,v 1.11 2023/02/02 08:21:32 mlelstv Exp $");
37 #if 0
38 __FBSDID("$FreeBSD: head/sbin/nvmecontrol/logpage.c 329824 2018-02-22 13:32:31Z wma $");
39 #endif
40 #endif
41 
42 #include <sys/param.h>
43 #include <sys/ioccom.h>
44 #include <sys/endian.h>
45 
46 #include <ctype.h>
47 #include <err.h>
48 #include <fcntl.h>
49 #include <stdbool.h>
50 #include <stddef.h>
51 #include <stdio.h>
52 #include <stdlib.h>
53 #include <string.h>
54 #include <unistd.h>
55 
56 #include "nvmectl.h"
57 
58 #define DEFAULT_SIZE	(4096)
59 #define MAX_FW_SLOTS	(7)
60 
61 typedef void (*print_fn_t)(const struct nvm_identify_controller *cdata, void *buf,
62     uint32_t size);
63 
64 struct kv_name {
65 	uint32_t key;
66 	const char *name;
67 };
68 
69 static const char *
kv_lookup(const struct kv_name * kv,size_t kv_count,uint32_t key)70 kv_lookup(const struct kv_name *kv, size_t kv_count, uint32_t key)
71 {
72 	static char bad[32];
73 	size_t i;
74 
75 	for (i = 0; i < kv_count; i++, kv++)
76 		if (kv->key == key)
77 			return kv->name;
78 	snprintf(bad, sizeof(bad), "Attribute %#x", key);
79 	return bad;
80 }
81 
82 static void
print_log_hex(const struct nvm_identify_controller * cdata __unused,void * data,uint32_t length)83 print_log_hex(const struct nvm_identify_controller *cdata __unused, void *data,
84     uint32_t length)
85 {
86 	print_hex(data, length);
87 }
88 
89 static void
print_bin(const struct nvm_identify_controller * cdata __unused,void * data,uint32_t length)90 print_bin(const struct nvm_identify_controller *cdata __unused, void *data,
91     uint32_t length)
92 {
93 	write(STDOUT_FILENO, data, length);
94 }
95 
96 static void *
get_log_buffer(uint32_t size)97 get_log_buffer(uint32_t size)
98 {
99 	void	*buf;
100 
101 	if ((buf = malloc(size)) == NULL)
102 		errx(1, "unable to malloc %u bytes", size);
103 
104 	memset(buf, 0, size);
105 	return (buf);
106 }
107 
108 void
read_logpage(int fd,uint8_t log_page,int nsid,void * payload,uint32_t payload_size)109 read_logpage(int fd, uint8_t log_page, int nsid, void *payload,
110     uint32_t payload_size)
111 {
112 	struct nvme_pt_command	pt;
113 
114 	memset(&pt, 0, sizeof(pt));
115 	pt.cmd.opcode = NVM_ADMIN_GET_LOG_PG;
116 	pt.cmd.nsid = nsid;
117 	pt.cmd.cdw10 = ((payload_size/sizeof(uint32_t)) - 1) << 16;
118 	pt.cmd.cdw10 |= log_page;
119 	pt.buf = payload;
120 	pt.len = payload_size;
121 	pt.is_read = 1;
122 
123 	if (ioctl(fd, NVME_PASSTHROUGH_CMD, &pt) < 0)
124 		err(1, "get log page request failed");
125 
126 	if (nvme_completion_is_error(&pt.cpl))
127 		errx(1, "get log page request returned error 0x%x/0x%02x",
128 			(uint8_t)__SHIFTOUT(pt.cpl.flags, NVME_CQE_SCT_MASK),
129 			(uint8_t)__SHIFTOUT(pt.cpl.flags, NVME_CQE_SC_MASK));
130 }
131 
132 static void
nvme_error_information_entry_swapbytes(struct nvme_error_information_entry * e)133 nvme_error_information_entry_swapbytes(struct nvme_error_information_entry *e)
134 {
135 #if _BYTE_ORDER != _LITTLE_ENDIAN
136 	e->error_count = le64toh(e->error_count);
137 	e->sqid = le16toh(e->sqid);
138 	e->cid = le16toh(e->cid);
139 	e->status = le16toh(e->status);
140 	e->error_location = le16toh(e->error_location);
141 	e->lba = le64toh(e->lba);
142 	e->nsid = le32toh(e->nsid);
143 	e->command_specific = le64toh(e->command_specific);
144 #endif
145 }
146 
147 static void
print_log_error(const struct nvm_identify_controller * cdata __unused,void * buf,uint32_t size)148 print_log_error(const struct nvm_identify_controller *cdata __unused, void *buf,
149     uint32_t size)
150 {
151 	int					i, nentries;
152 	struct nvme_error_information_entry	*entry = buf;
153 
154 	/* Convert data to host endian */
155 	nvme_error_information_entry_swapbytes(entry);
156 
157 	printf("Error Information Log\n");
158 	printf("=====================\n");
159 
160 	if (entry->error_count == 0) {
161 		printf("No error entries found\n");
162 		return;
163 	}
164 
165 	nentries = size/sizeof(struct nvme_error_information_entry);
166 	for (i = 0; i < nentries; i++, entry++) {
167 		if (entry->error_count == 0)
168 			break;
169 
170 		printf("Entry %02d\n", i + 1);
171 		printf("=========\n");
172 		printf(" Error count:           %ju\n", entry->error_count);
173 		printf(" Submission queue ID:   %u\n", entry->sqid);
174 		printf(" Command ID:            %u\n", entry->cid);
175 		/* TODO: Export nvme_status_string structures from kernel? */
176 		printf(" Status:\n");
177 		printf("  Phase tag:            %d\n",
178 		    (uint16_t)__SHIFTOUT(entry->status, NVME_CQE_PHASE));
179 		printf("  Status code:          %d\n",
180 		    (uint16_t)__SHIFTOUT(entry->status, NVME_CQE_SC_MASK));
181 		printf("  Status code type:     %d\n",
182 		    (uint16_t)__SHIFTOUT(entry->status, NVME_CQE_SCT_MASK));
183 		printf("  More:                 %d\n",
184 		    (uint16_t)__SHIFTOUT(entry->status, NVME_CQE_M));
185 		printf("  DNR:                  %d\n",
186 		    (uint16_t)__SHIFTOUT(entry->status, NVME_CQE_DNR));
187 		printf(" Error location:        %u\n", entry->error_location);
188 		printf(" LBA:                   %ju\n", entry->lba);
189 		printf(" Namespace ID:          %u\n", entry->nsid);
190 		printf(" Vendor specific info:  %u\n", entry->vendor_specific);
191 		printf(" Command specific info: %ju\n",
192 		    entry->command_specific);
193 	}
194 }
195 
196 static void
print_temp(uint16_t t)197 print_temp(uint16_t t)
198 {
199 	printf("%u K, %2.2f C, %3.2f F\n", t, (float)t - 273.15,
200 	    (float)t * 9 / 5 - 459.67);
201 }
202 
203 static void
nvme_health_information_page_swapbytes(struct nvme_health_information_page * e)204 nvme_health_information_page_swapbytes(struct nvme_health_information_page *e)
205 {
206 #if _BYTE_ORDER != _LITTLE_ENDIAN
207 	u_int i;
208 
209 	e->composite_temperature = le16toh(e->composite_temperature);
210 	nvme_le128toh(e->data_units_read);
211 	nvme_le128toh(e->data_units_written);
212 	nvme_le128toh(e->host_read_commands);
213 	nvme_le128toh(e->host_write_commands);
214 	nvme_le128toh(e->controller_busy_time);
215 	nvme_le128toh(e->power_cycles);
216 	nvme_le128toh(e->power_on_hours);
217 	nvme_le128toh(e->unsafe_shutdowns);
218 	nvme_le128toh(e->media_errors);
219 	nvme_le128toh(e->num_error_info_log_entries);
220 	e->warning_temp_time = le32toh(e->warning_temp_time);
221 	e->error_temp_time = le32toh(e->error_temp_time);
222 	for (i = 0; i < __arraycount(e->temp_sensor); i++)
223 		e->temp_sensor[i] = le16toh(e->temp_sensor[i]);
224 #endif
225 }
226 
227 static void
print_log_health(const struct nvm_identify_controller * cdata __unused,void * buf,uint32_t size __unused)228 print_log_health(const struct nvm_identify_controller *cdata __unused, void *buf,
229     uint32_t size __unused)
230 {
231 	struct nvme_health_information_page *health = buf;
232 	u_int i;
233 
234 	/* Convert data to host endian */
235 	nvme_health_information_page_swapbytes(health);
236 
237 	printf("SMART/Health Information Log\n");
238 	printf("============================\n");
239 
240 	printf("Critical Warning State:         0x%02x\n",
241 	    health->critical_warning);
242 	printf(" Available spare:               %d\n",
243 	    (uint8_t)__SHIFTOUT(health->critical_warning,
244 	      NVME_HEALTH_PAGE_CW_AVAIL_SPARE));
245 	printf(" Temperature:                   %d\n",
246 	    (uint8_t)__SHIFTOUT(health->critical_warning,
247 	      NVME_HEALTH_PAGE_CW_TEMPERTURE));
248 	printf(" Device reliability:            %d\n",
249 	    (uint8_t)__SHIFTOUT(health->critical_warning,
250 	      NVME_HEALTH_PAGE_CW_DEVICE_RELIABLITY));
251 	printf(" Read only:                     %d\n",
252 	    (uint8_t)__SHIFTOUT(health->critical_warning,
253 	      NVME_HEALTH_PAGE_CW_READ_ONLY));
254 	printf(" Volatile memory backup:        %d\n",
255 	    (uint8_t)__SHIFTOUT(health->critical_warning,
256 	      NVME_HEALTH_PAGE_CW_VOLATILE_MEMORY_BACKUP));
257 	printf("Temperature:                    ");
258 	print_temp(health->composite_temperature);
259 	printf("Available spare:                %u\n",
260 	    health->available_spare);
261 	printf("Available spare threshold:      %u\n",
262 	    health->available_spare_threshold);
263 	printf("Percentage used:                %u\n",
264 	    health->percentage_used);
265 
266 	print_bignum1("Data units read:", health->data_units_read, "", "B", 512000);
267 	print_bignum1("Data units written:", health->data_units_written,
268 	    "", "B", 512000);
269 	print_bignum("Host read commands:", health->host_read_commands, "");
270 	print_bignum("Host write commands:", health->host_write_commands, "");
271 	print_bignum("Controller busy time (minutes):", health->controller_busy_time,
272 	    "");
273 	print_bignum("Power cycles:", health->power_cycles, "");
274 	print_bignum("Power on hours:", health->power_on_hours, "");
275 	print_bignum("Unsafe shutdowns:", health->unsafe_shutdowns, "");
276 	print_bignum("Media errors:", health->media_errors, "");
277 	print_bignum("No. error info log entries:",
278 	    health->num_error_info_log_entries, "");
279 
280 	printf("Warning Temp Composite Time:    %d\n", health->warning_temp_time);
281 	printf("Error Temp Composite Time:      %d\n", health->error_temp_time);
282 	for (i = 0; i < __arraycount(health->temp_sensor); i++) {
283 		if (health->temp_sensor[i] == 0)
284 			continue;
285 		printf("Temperature Sensor %d:           ", i + 1);
286 		print_temp(health->temp_sensor[i]);
287 	}
288 }
289 
290 static void
nvme_firmware_page_swapbytes(struct nvme_firmware_page * e)291 nvme_firmware_page_swapbytes(struct nvme_firmware_page *e)
292 {
293 #if _BYTE_ORDER != _LITTLE_ENDIAN
294 	u_int i;
295 
296 	for (i = 0; i < __arraycount(e->revision); i++)
297 		e->revision[i] = le64toh(e->revision[i]);
298 #endif
299 }
300 
301 static void
print_log_firmware(const struct nvm_identify_controller * cdata,void * buf,uint32_t size __unused)302 print_log_firmware(const struct nvm_identify_controller *cdata, void *buf,
303     uint32_t size __unused)
304 {
305 	u_int				i, slots;
306 	const char			*status;
307 	struct nvme_firmware_page	*fw = buf;
308 
309 	/* Convert data to host endian */
310 	nvme_firmware_page_swapbytes(fw);
311 
312 	printf("Firmware Slot Log\n");
313 	printf("=================\n");
314 
315 	if (!(cdata->oacs & NVME_ID_CTRLR_OACS_FW))
316 		slots = 1;
317 	else
318 		slots = MIN(__SHIFTOUT(cdata->frmw, NVME_ID_CTRLR_FRMW_NSLOT),
319 		    MAX_FW_SLOTS);
320 
321 	for (i = 0; i < slots; i++) {
322 		printf("Slot %d: ", i + 1);
323 		if (__SHIFTOUT(fw->afi, NVME_FW_PAGE_AFI_SLOT) == i + 1)
324 			status = "  Active";
325 		else
326 			status = "Inactive";
327 
328 		if (fw->revision[i] == 0LLU)
329 			printf("Empty\n");
330 		else
331 			if (isprint(*(uint8_t *)&fw->revision[i]))
332 				printf("[%s] %.8s\n", status,
333 				    (char *)&fw->revision[i]);
334 			else
335 				printf("[%s] %016jx\n", status,
336 				    fw->revision[i]);
337 	}
338 }
339 
340 /*
341  * Intel specific log pages from
342  * http://www.intel.com/content/dam/www/public/us/en/documents/product-specifications/ssd-dc-p3700-spec.pdf
343  *
344  * Though the version as of this date has a typo for the size of log page 0xca,
345  * offset 147: it is only 1 byte, not 6.
346  */
347 static void
intel_log_temp_stats_swapbytes(struct intel_log_temp_stats * e)348 intel_log_temp_stats_swapbytes(struct intel_log_temp_stats *e)
349 {
350 #if _BYTE_ORDER != _LITTLE_ENDIAN
351 	e->current = le64toh(e->current);
352 	e->overtemp_flag_last = le64toh(e->overtemp_flag_last);
353 	e->overtemp_flag_life = le64toh(e->overtemp_flag_life);
354 	e->max_temp = le64toh(e->max_temp);
355 	e->min_temp = le64toh(e->min_temp);
356 	e->max_oper_temp = le64toh(e->max_oper_temp);
357 	e->min_oper_temp = le64toh(e->min_oper_temp);
358 	e->est_offset = le64toh(e->est_offset);
359 #endif
360 }
361 
362 static void
print_intel_temp_stats(const struct nvm_identify_controller * cdata __unused,void * buf,uint32_t size __unused)363 print_intel_temp_stats(const struct nvm_identify_controller *cdata __unused,
364     void *buf, uint32_t size __unused)
365 {
366 	struct intel_log_temp_stats	*temp = buf;
367 
368 	/* Convert data to host endian */
369 	intel_log_temp_stats_swapbytes(temp);
370 
371 	printf("Intel Temperature Log\n");
372 	printf("=====================\n");
373 
374 	printf("Current:                        ");
375 	print_temp(temp->current);
376 	printf("Overtemp Last Flags             %#jx\n",
377 	    (uintmax_t)temp->overtemp_flag_last);
378 	printf("Overtemp Lifetime Flags         %#jx\n",
379 	    (uintmax_t)temp->overtemp_flag_life);
380 	printf("Max Temperature                 ");
381 	print_temp(temp->max_temp);
382 	printf("Min Temperature                 ");
383 	print_temp(temp->min_temp);
384 	printf("Max Operating Temperature       ");
385 	print_temp(temp->max_oper_temp);
386 	printf("Min Operating Temperature       ");
387 	print_temp(temp->min_oper_temp);
388 	printf("Estimated Temperature Offset:   %ju C/K\n",
389 	    (uintmax_t)temp->est_offset);
390 }
391 
392 /*
393  * Format from Table 22, section 5.7 IO Command Latency Statistics.
394  * Read and write stats pages have identical encoding.
395  */
396 static void
print_intel_read_write_lat_log(const struct nvm_identify_controller * cdata __unused,void * buf,uint32_t size __unused)397 print_intel_read_write_lat_log(const struct nvm_identify_controller *cdata __unused,
398     void *buf, uint32_t size __unused)
399 {
400 	const char *walker = buf;
401 	int i;
402 
403 	printf("Major:                         %d\n", le16dec(walker + 0));
404 	printf("Minor:                         %d\n", le16dec(walker + 2));
405 	for (i = 0; i < 32; i++)
406 		printf("%4dus-%4dus:                 %ju\n", i * 32, (i + 1) * 32,
407 		    (uintmax_t)le32dec(walker + 4 + i * 4));
408 	for (i = 1; i < 32; i++)
409 		printf("%4dms-%4dms:                 %ju\n", i, i + 1,
410 		    (uintmax_t)le32dec(walker + 132 + i * 4));
411 	for (i = 1; i < 32; i++)
412 		printf("%4dms-%4dms:                 %ju\n", i * 32, (i + 1) * 32,
413 		    (uintmax_t)le32dec(walker + 256 + i * 4));
414 }
415 
416 static void
print_intel_read_lat_log(const struct nvm_identify_controller * cdata,void * buf,uint32_t size)417 print_intel_read_lat_log(const struct nvm_identify_controller *cdata, void *buf,
418     uint32_t size)
419 {
420 
421 	printf("Intel Read Latency Log\n");
422 	printf("======================\n");
423 	print_intel_read_write_lat_log(cdata, buf, size);
424 }
425 
426 static void
print_intel_write_lat_log(const struct nvm_identify_controller * cdata,void * buf,uint32_t size)427 print_intel_write_lat_log(const struct nvm_identify_controller *cdata, void *buf,
428     uint32_t size)
429 {
430 
431 	printf("Intel Write Latency Log\n");
432 	printf("=======================\n");
433 	print_intel_read_write_lat_log(cdata, buf, size);
434 }
435 
436 /*
437  * Table 19. 5.4 SMART Attributes.
438  * Samsung also implements this and some extra data not documented.
439  */
440 static void
print_intel_add_smart(const struct nvm_identify_controller * cdata __unused,void * buf,uint32_t size __unused)441 print_intel_add_smart(const struct nvm_identify_controller *cdata __unused,
442     void *buf, uint32_t size __unused)
443 {
444 	uint8_t *walker = buf;
445 	uint8_t *end = walker + 150;
446 	const char *name;
447 	uint64_t raw;
448 	uint8_t normalized;
449 
450 	static struct kv_name kv[] = {
451 		{ 0xab, "Program Fail Count" },
452 		{ 0xac, "Erase Fail Count" },
453 		{ 0xad, "Wear Leveling Count" },
454 		{ 0xb8, "End to End Error Count" },
455 		{ 0xc7, "CRC Error Count" },
456 		{ 0xe2, "Timed: Media Wear" },
457 		{ 0xe3, "Timed: Host Read %" },
458 		{ 0xe4, "Timed: Elapsed Time" },
459 		{ 0xea, "Thermal Throttle Status" },
460 		{ 0xf0, "Retry Buffer Overflows" },
461 		{ 0xf3, "PLL Lock Loss Count" },
462 		{ 0xf4, "NAND Bytes Written" },
463 		{ 0xf5, "Host Bytes Written" },
464 	};
465 
466 	printf("Additional SMART Data Log\n");
467 	printf("=========================\n");
468 	/*
469 	 * walker[0] = Key
470 	 * walker[1,2] = reserved
471 	 * walker[3] = Normalized Value
472 	 * walker[4] = reserved
473 	 * walker[5..10] = Little Endian Raw value
474 	 *	(or other represenations)
475 	 * walker[11] = reserved
476 	 */
477 	while (walker < end) {
478 		name = kv_lookup(kv, __arraycount(kv), *walker);
479 		normalized = walker[3];
480 		raw = le48dec(walker + 5);
481 		switch (*walker){
482 		case 0:
483 			break;
484 		case 0xad:
485 			printf("%-32s: %3d min: %u max: %u ave: %u\n", name,
486 			    normalized, le16dec(walker + 5), le16dec(walker + 7),
487 			    le16dec(walker + 9));
488 			break;
489 		case 0xe2:
490 			printf("%-32s: %3d %.3f%%\n", name, normalized, raw / 1024.0);
491 			break;
492 		case 0xea:
493 			printf("%-32s: %3d %d%% %d times\n", name, normalized,
494 			    walker[5], le32dec(walker+6));
495 			break;
496 		default:
497 			printf("%-32s: %3d %ju\n", name, normalized, (uintmax_t)raw);
498 			break;
499 		}
500 		walker += 12;
501 	}
502 }
503 
504 /*
505  * HGST's 0xc1 page. This is a grab bag of additional data. Please see
506  * https://www.hgst.com/sites/default/files/resources/US_SN150_ProdManual.pdf
507  * https://www.hgst.com/sites/default/files/resources/US_SN100_ProdManual.pdf
508  * Appendix A for details
509  */
510 
511 typedef void (*subprint_fn_t)(void *buf, uint16_t subtype, uint8_t res,
512     uint32_t size);
513 
514 struct subpage_print {
515 	uint16_t key;
516 	subprint_fn_t fn;
517 };
518 
519 static void print_hgst_info_write_errors(void *, uint16_t, uint8_t, uint32_t);
520 static void print_hgst_info_read_errors(void *, uint16_t, uint8_t, uint32_t);
521 static void print_hgst_info_verify_errors(void *, uint16_t, uint8_t, uint32_t);
522 static void print_hgst_info_self_test(void *, uint16_t, uint8_t, uint32_t);
523 static void print_hgst_info_background_scan(void *, uint16_t, uint8_t, uint32_t);
524 static void print_hgst_info_erase_errors(void *, uint16_t, uint8_t, uint32_t);
525 static void print_hgst_info_erase_counts(void *, uint16_t, uint8_t, uint32_t);
526 static void print_hgst_info_temp_history(void *, uint16_t, uint8_t, uint32_t);
527 static void print_hgst_info_ssd_perf(void *, uint16_t, uint8_t, uint32_t);
528 static void print_hgst_info_firmware_load(void *, uint16_t, uint8_t, uint32_t);
529 
530 static struct subpage_print hgst_subpage[] = {
531 	{ 0x02, print_hgst_info_write_errors },
532 	{ 0x03, print_hgst_info_read_errors },
533 	{ 0x05, print_hgst_info_verify_errors },
534 	{ 0x10, print_hgst_info_self_test },
535 	{ 0x15, print_hgst_info_background_scan },
536 	{ 0x30, print_hgst_info_erase_errors },
537 	{ 0x31, print_hgst_info_erase_counts },
538 	{ 0x32, print_hgst_info_temp_history },
539 	{ 0x37, print_hgst_info_ssd_perf },
540 	{ 0x38, print_hgst_info_firmware_load },
541 };
542 
543 /* Print a subpage that is basically just key value pairs */
544 static void
print_hgst_info_subpage_gen(void * buf,uint16_t subtype __unused,uint32_t size,const struct kv_name * kv,size_t kv_count)545 print_hgst_info_subpage_gen(void *buf, uint16_t subtype __unused, uint32_t size,
546     const struct kv_name *kv, size_t kv_count)
547 {
548 	uint8_t *wsp, *esp;
549 	uint16_t ptype;
550 	uint8_t plen;
551 	uint64_t param;
552 	int i;
553 
554 	wsp = buf;
555 	esp = wsp + size;
556 	while (wsp < esp) {
557 		ptype = le16dec(wsp);
558 		wsp += 2;
559 		wsp++;			/* Flags, just ignore */
560 		plen = *wsp++;
561 		param = 0;
562 		for (i = 0; i < plen; i++)
563 			param |= (uint64_t)*wsp++ << (i * 8);
564 		printf("  %-30s: %jd\n", kv_lookup(kv, kv_count, ptype),
565 		    (uintmax_t)param);
566 	}
567 }
568 
569 static void
print_hgst_info_write_errors(void * buf,uint16_t subtype,uint8_t res __unused,uint32_t size)570 print_hgst_info_write_errors(void *buf, uint16_t subtype, uint8_t res __unused,
571     uint32_t size)
572 {
573 	static const struct kv_name kv[] = {
574 		{ 0x0000, "Corrected Without Delay" },
575 		{ 0x0001, "Corrected Maybe Delayed" },
576 		{ 0x0002, "Re-Writes" },
577 		{ 0x0003, "Errors Corrected" },
578 		{ 0x0004, "Correct Algorithm Used" },
579 		{ 0x0005, "Bytes Processed" },
580 		{ 0x0006, "Uncorrected Errors" },
581 		{ 0x8000, "Flash Write Commands" },
582 		{ 0x8001, "HGST Special" },
583 	};
584 
585 	printf("Write Errors Subpage:\n");
586 	print_hgst_info_subpage_gen(buf, subtype, size, kv, __arraycount(kv));
587 }
588 
589 static void
print_hgst_info_read_errors(void * buf,uint16_t subtype,uint8_t res __unused,uint32_t size)590 print_hgst_info_read_errors(void *buf, uint16_t subtype, uint8_t res __unused,
591     uint32_t size)
592 {
593 	static const struct kv_name kv[] = {
594 		{ 0x0000, "Corrected Without Delay" },
595 		{ 0x0001, "Corrected Maybe Delayed" },
596 		{ 0x0002, "Re-Reads" },
597 		{ 0x0003, "Errors Corrected" },
598 		{ 0x0004, "Correct Algorithm Used" },
599 		{ 0x0005, "Bytes Processed" },
600 		{ 0x0006, "Uncorrected Errors" },
601 		{ 0x8000, "Flash Read Commands" },
602 		{ 0x8001, "XOR Recovered" },
603 		{ 0x8002, "Total Corrected Bits" },
604 	};
605 
606 	printf("Read Errors Subpage:\n");
607 	print_hgst_info_subpage_gen(buf, subtype, size, kv, __arraycount(kv));
608 }
609 
610 static void
print_hgst_info_verify_errors(void * buf,uint16_t subtype,uint8_t res __unused,uint32_t size)611 print_hgst_info_verify_errors(void *buf, uint16_t subtype, uint8_t res __unused,
612     uint32_t size)
613 {
614 	static const struct kv_name kv[] = {
615 		{ 0x0000, "Corrected Without Delay" },
616 		{ 0x0001, "Corrected Maybe Delayed" },
617 		{ 0x0002, "Re-Reads" },
618 		{ 0x0003, "Errors Corrected" },
619 		{ 0x0004, "Correct Algorithm Used" },
620 		{ 0x0005, "Bytes Processed" },
621 		{ 0x0006, "Uncorrected Errors" },
622 		{ 0x8000, "Commands Processed" },
623 	};
624 
625 	printf("Verify Errors Subpage:\n");
626 	print_hgst_info_subpage_gen(buf, subtype, size, kv, __arraycount(kv));
627 }
628 
629 static void
print_hgst_info_self_test(void * buf,uint16_t subtype __unused,uint8_t res __unused,uint32_t size)630 print_hgst_info_self_test(void *buf, uint16_t subtype __unused, uint8_t res __unused,
631     uint32_t size)
632 {
633 	size_t i;
634 	uint8_t *walker = buf;
635 	uint16_t code, hrs;
636 	uint32_t lba;
637 
638 	printf("Self Test Subpage:\n");
639 	for (i = 0; i < size / 20; i++) {	/* Each entry is 20 bytes */
640 		code = le16dec(walker);
641 		walker += 2;
642 		walker++;			/* Ignore fixed flags */
643 		if (*walker == 0)		/* Last entry is zero length */
644 			break;
645 		if (*walker++ != 0x10) {
646 			printf("Bad length for self test report\n");
647 			return;
648 		}
649 		printf("  %-30s: %d\n", "Recent Test", code);
650 		printf("    %-28s: %#x\n", "Self-Test Results", *walker & 0xf);
651 		printf("    %-28s: %#x\n", "Self-Test Code", (*walker >> 5) & 0x7);
652 		walker++;
653 		printf("    %-28s: %#x\n", "Self-Test Number", *walker++);
654 		hrs = le16dec(walker);
655 		walker += 2;
656 		lba = le32dec(walker);
657 		walker += 4;
658 		printf("    %-28s: %u\n", "Total Power On Hrs", hrs);
659 		printf("    %-28s: %#jx (%jd)\n", "LBA", (uintmax_t)lba,
660 		    (uintmax_t)lba);
661 		printf("    %-28s: %#x\n", "Sense Key", *walker++ & 0xf);
662 		printf("    %-28s: %#x\n", "Additional Sense Code", *walker++);
663 		printf("    %-28s: %#x\n", "Additional Sense Qualifier", *walker++);
664 		printf("    %-28s: %#x\n", "Vendor Specific Detail", *walker++);
665 	}
666 }
667 
668 static void
print_hgst_info_background_scan(void * buf,uint16_t subtype __unused,uint8_t res __unused,uint32_t size)669 print_hgst_info_background_scan(void *buf, uint16_t subtype __unused,
670     uint8_t res __unused, uint32_t size)
671 {
672 	uint8_t *walker = buf;
673 	uint8_t status;
674 	uint16_t code, nscan, progress;
675 	uint32_t pom, nand;
676 
677 	printf("Background Media Scan Subpage:\n");
678 	/* Decode the header */
679 	code = le16dec(walker);
680 	walker += 2;
681 	walker++;			/* Ignore fixed flags */
682 	if (*walker++ != 0x10) {
683 		printf("Bad length for background scan header\n");
684 		return;
685 	}
686 	if (code != 0) {
687 		printf("Expected code 0, found code %#x\n", code);
688 		return;
689 	}
690 	pom = le32dec(walker);
691 	walker += 4;
692 	walker++;			/* Reserved */
693 	status = *walker++;
694 	nscan = le16dec(walker);
695 	walker += 2;
696 	progress = le16dec(walker);
697 	walker += 2;
698 	walker += 6;			/* Reserved */
699 	printf("  %-30s: %d\n", "Power On Minutes", pom);
700 	printf("  %-30s: %x (%s)\n", "BMS Status", status,
701 	    status == 0 ? "idle" : (status == 1 ? "active" :
702 	      (status == 8 ? "suspended" : "unknown")));
703 	printf("  %-30s: %d\n", "Number of BMS", nscan);
704 	printf("  %-30s: %d\n", "Progress Current BMS", progress);
705 	/* Report retirements */
706 	if (walker - (uint8_t *)buf != 20) {
707 		printf("Coding error, offset not 20\n");
708 		return;
709 	}
710 	size -= 20;
711 	printf("  %-30s: %d\n", "BMS retirements", size / 0x18);
712 	while (size > 0) {
713 		code = le16dec(walker);
714 		walker += 2;
715 		walker++;
716 		if (*walker++ != 0x14) {
717 			printf("Bad length parameter\n");
718 			return;
719 		}
720 		pom = le32dec(walker);
721 		walker += 4;
722 		/*
723 		 * Spec sheet says the following are hard coded, if true, just
724 		 * print the NAND retirement.
725 		 */
726 		if (walker[0] == 0x41 &&
727 		    walker[1] == 0x0b &&
728 		    walker[2] == 0x01 &&
729 		    walker[3] == 0x00 &&
730 		    walker[4] == 0x00 &&
731 		    walker[5] == 0x00 &&
732 		    walker[6] == 0x00 &&
733 		    walker[7] == 0x00) {
734 			walker += 8;
735 			walker += 4;	/* Skip reserved */
736 			nand = le32dec(walker);
737 			walker += 4;
738 			printf("  %-30s: %d\n", "Retirement number", code);
739 			printf("    %-28s: %#x\n", "NAND (C/T)BBBPPP", nand);
740 		} else {
741 			printf("Parameter %#x entry corrupt\n", code);
742 			walker += 16;
743 		}
744 	}
745 }
746 
747 static void
print_hgst_info_erase_errors(void * buf,uint16_t subtype __unused,uint8_t res __unused,uint32_t size)748 print_hgst_info_erase_errors(void *buf, uint16_t subtype __unused,
749     uint8_t res __unused, uint32_t size)
750 {
751 	static const struct kv_name kv[] = {
752 		{ 0x0000, "Corrected Without Delay" },
753 		{ 0x0001, "Corrected Maybe Delayed" },
754 		{ 0x0002, "Re-Erase" },
755 		{ 0x0003, "Errors Corrected" },
756 		{ 0x0004, "Correct Algorithm Used" },
757 		{ 0x0005, "Bytes Processed" },
758 		{ 0x0006, "Uncorrected Errors" },
759 		{ 0x8000, "Flash Erase Commands" },
760 		{ 0x8001, "Mfg Defect Count" },
761 		{ 0x8002, "Grown Defect Count" },
762 		{ 0x8003, "Erase Count -- User" },
763 		{ 0x8004, "Erase Count -- System" },
764 	};
765 
766 	printf("Erase Errors Subpage:\n");
767 	print_hgst_info_subpage_gen(buf, subtype, size, kv, __arraycount(kv));
768 }
769 
770 static void
print_hgst_info_erase_counts(void * buf,uint16_t subtype,uint8_t res __unused,uint32_t size)771 print_hgst_info_erase_counts(void *buf, uint16_t subtype, uint8_t res __unused,
772     uint32_t size)
773 {
774 	/* My drive doesn't export this -- so not coding up */
775 	printf("XXX: Erase counts subpage: %p, %#x %d\n", buf, subtype, size);
776 }
777 
778 static void
print_hgst_info_temp_history(void * buf,uint16_t subtype __unused,uint8_t res __unused,uint32_t size __unused)779 print_hgst_info_temp_history(void *buf, uint16_t subtype __unused,
780     uint8_t res __unused, uint32_t size __unused)
781 {
782 	uint8_t *walker = buf;
783 	uint32_t min;
784 
785 	printf("Temperature History:\n");
786 	printf("  %-30s: %d C\n", "Current Temperature", *walker++);
787 	printf("  %-30s: %d C\n", "Reference Temperature", *walker++);
788 	printf("  %-30s: %d C\n", "Maximum Temperature", *walker++);
789 	printf("  %-30s: %d C\n", "Minimum Temperature", *walker++);
790 	min = le32dec(walker);
791 	walker += 4;
792 	printf("  %-30s: %d:%02d:00\n", "Max Temperature Time", min / 60, min % 60);
793 	min = le32dec(walker);
794 	walker += 4;
795 	printf("  %-30s: %d:%02d:00\n", "Over Temperature Duration", min / 60,
796 	    min % 60);
797 	min = le32dec(walker);
798 	walker += 4;
799 	printf("  %-30s: %d:%02d:00\n", "Min Temperature Time", min / 60, min % 60);
800 }
801 
802 static void
print_hgst_info_ssd_perf(void * buf,uint16_t subtype __unused,uint8_t res,uint32_t size __unused)803 print_hgst_info_ssd_perf(void *buf, uint16_t subtype __unused, uint8_t res,
804     uint32_t size __unused)
805 {
806 	uint8_t *walker = buf;
807 	uint64_t val;
808 
809 	printf("SSD Performance Subpage Type %d:\n", res);
810 	val = le64dec(walker);
811 	walker += 8;
812 	printf("  %-30s: %ju\n", "Host Read Commands", val);
813 	val = le64dec(walker);
814 	walker += 8;
815 	printf("  %-30s: %ju\n", "Host Read Blocks", val);
816 	val = le64dec(walker);
817 	walker += 8;
818 	printf("  %-30s: %ju\n", "Host Cache Read Hits Commands", val);
819 	val = le64dec(walker);
820 	walker += 8;
821 	printf("  %-30s: %ju\n", "Host Cache Read Hits Blocks", val);
822 	val = le64dec(walker);
823 	walker += 8;
824 	printf("  %-30s: %ju\n", "Host Read Commands Stalled", val);
825 	val = le64dec(walker);
826 	walker += 8;
827 	printf("  %-30s: %ju\n", "Host Write Commands", val);
828 	val = le64dec(walker);
829 	walker += 8;
830 	printf("  %-30s: %ju\n", "Host Write Blocks", val);
831 	val = le64dec(walker);
832 	walker += 8;
833 	printf("  %-30s: %ju\n", "Host Write Odd Start Commands", val);
834 	val = le64dec(walker);
835 	walker += 8;
836 	printf("  %-30s: %ju\n", "Host Write Odd End Commands", val);
837 	val = le64dec(walker);
838 	walker += 8;
839 	printf("  %-30s: %ju\n", "Host Write Commands Stalled", val);
840 	val = le64dec(walker);
841 	walker += 8;
842 	printf("  %-30s: %ju\n", "NAND Read Commands", val);
843 	val = le64dec(walker);
844 	walker += 8;
845 	printf("  %-30s: %ju\n", "NAND Read Blocks", val);
846 	val = le64dec(walker);
847 	walker += 8;
848 	printf("  %-30s: %ju\n", "NAND Write Commands", val);
849 	val = le64dec(walker);
850 	walker += 8;
851 	printf("  %-30s: %ju\n", "NAND Write Blocks", val);
852 	val = le64dec(walker);
853 	walker += 8;
854 	printf("  %-30s: %ju\n", "NAND Read Before Writes", val);
855 }
856 
857 static void
print_hgst_info_firmware_load(void * buf,uint16_t subtype __unused,uint8_t res __unused,uint32_t size __unused)858 print_hgst_info_firmware_load(void *buf, uint16_t subtype __unused,
859     uint8_t res __unused, uint32_t size __unused)
860 {
861 	uint8_t *walker = buf;
862 
863 	printf("Firmware Load Subpage:\n");
864 	printf("  %-30s: %d\n", "Firmware Downloads", le32dec(walker));
865 }
866 
867 static void
kv_indirect(void * buf,uint32_t subtype,uint8_t res,uint32_t size,struct subpage_print * sp,size_t nsp)868 kv_indirect(void *buf, uint32_t subtype, uint8_t res, uint32_t size,
869     struct subpage_print *sp, size_t nsp)
870 {
871 	size_t i;
872 
873 	for (i = 0; i < nsp; i++, sp++) {
874 		if (sp->key == subtype) {
875 			sp->fn(buf, subtype, res, size);
876 			return;
877 		}
878 	}
879 	printf("No handler for page type %x\n", subtype);
880 }
881 
882 static void
print_hgst_info_log(const struct nvm_identify_controller * cdata __unused,void * buf,uint32_t size __unused)883 print_hgst_info_log(const struct nvm_identify_controller *cdata __unused, void *buf,
884     uint32_t size __unused)
885 {
886 	uint8_t	*walker, *end, *subpage;
887 	int pages __unused;
888 	uint16_t len;
889 	uint8_t subtype, res;
890 
891 	printf("HGST Extra Info Log\n");
892 	printf("===================\n");
893 
894 	walker = buf;
895 	pages = *walker++;
896 	walker++;
897 	len = le16dec(walker);
898 	walker += 2;
899 	end = walker + len;		/* Length is exclusive of this header */
900 
901 	while (walker < end) {
902 		subpage = walker + 4;
903 		subtype = *walker++ & 0x3f;	/* subtype */
904 		res = *walker++;		/* Reserved */
905 		len = le16dec(walker);
906 		walker += len + 2;		/* Length, not incl header */
907 		if (walker > end) {
908 			printf("Ooops! Off the end of the list\n");
909 			break;
910 		}
911 		kv_indirect(subpage, subtype, res, len, hgst_subpage,
912 		    __arraycount(hgst_subpage));
913 	}
914 }
915 
916 /*
917  * Table of log page printer / sizing.
918  *
919  * This includes Intel specific pages that are widely implemented.
920  * Make sure you keep all the pages of one vendor together so -v help
921  * lists all the vendors pages.
922  */
923 static struct logpage_function {
924 	uint8_t		log_page;
925 	const char     *vendor;
926 	const char     *name;
927 	print_fn_t	print_fn;
928 	size_t		size;
929 } logfuncs[] = {
930 	{NVME_LOG_ERROR,		NULL,	"Drive Error Log",
931 	 print_log_error,		0},
932 	{NVME_LOG_HEALTH_INFORMATION,	NULL,	"Health/SMART Data",
933 	 print_log_health,		sizeof(struct nvme_health_information_page)},
934 	{NVME_LOG_FIRMWARE_SLOT,	NULL,	"Firmware Information",
935 	 print_log_firmware,		sizeof(struct nvme_firmware_page)},
936 	{HGST_INFO_LOG,			"hgst",	"Detailed Health/SMART",
937 	 print_hgst_info_log,		DEFAULT_SIZE},
938 	{HGST_INFO_LOG,			"wds",	"Detailed Health/SMART",
939 	 print_hgst_info_log,		DEFAULT_SIZE},
940 	{INTEL_LOG_TEMP_STATS,		"intel", "Temperature Stats",
941 	 print_intel_temp_stats,	sizeof(struct intel_log_temp_stats)},
942 	{INTEL_LOG_READ_LAT_LOG,	"intel", "Read Latencies",
943 	 print_intel_read_lat_log,	DEFAULT_SIZE},
944 	{INTEL_LOG_WRITE_LAT_LOG,	"intel", "Write Latencies",
945 	 print_intel_write_lat_log,	DEFAULT_SIZE},
946 	{INTEL_LOG_ADD_SMART,		"intel", "Extra Health/SMART Data",
947 	 print_intel_add_smart,		DEFAULT_SIZE},
948 	{INTEL_LOG_ADD_SMART,		"samsung", "Extra Health/SMART Data",
949 	 print_intel_add_smart,		DEFAULT_SIZE},
950 
951 	{0, NULL, NULL, NULL, 0},
952 };
953 
954 __dead static void
logpage_usage(void)955 logpage_usage(void)
956 {
957 	fprintf(stderr, "usage:\n");
958 	fprintf(stderr, "\t%s " LOGPAGE_USAGE, getprogname());
959 	exit(1);
960 }
961 
962 __dead static void
logpage_help(void)963 logpage_help(void)
964 {
965 	struct logpage_function		*f;
966 	const char 			*v;
967 
968 	fprintf(stderr, "\n");
969 	fprintf(stderr, "%-8s %-10s %s\n", "Page", "Vendor","Page Name");
970 	fprintf(stderr, "-------- ---------- ----------\n");
971 	for (f = logfuncs; f->log_page > 0; f++) {
972 		v = f->vendor == NULL ? "-" : f->vendor;
973 		fprintf(stderr, "0x%02x     %-10s %s\n", f->log_page, v, f->name);
974 	}
975 
976 	exit(1);
977 }
978 
979 void
logpage(int argc,char * argv[])980 logpage(int argc, char *argv[])
981 {
982 	int				fd, nsid;
983 	int				log_page = 0, pageflag = false;
984 	int				binflag = false, hexflag = false, ns_specified;
985 	int				ch;
986 	char				*p;
987 	char				cname[64];
988 	uint32_t			size;
989 	void				*buf;
990 	const char 			*vendor = NULL;
991 	struct logpage_function		*f;
992 	struct nvm_identify_controller	cdata;
993 	print_fn_t			print_fn;
994 
995 	while ((ch = getopt(argc, argv, "bp:xv:")) != -1) {
996 		switch (ch) {
997 		case 'b':
998 			binflag = true;
999 			break;
1000 		case 'p':
1001 			if (strcmp(optarg, "help") == 0)
1002 				logpage_help();
1003 
1004 			/* TODO: Add human-readable ASCII page IDs */
1005 			log_page = strtol(optarg, &p, 0);
1006 			if (p != NULL && *p != '\0') {
1007 				fprintf(stderr,
1008 				    "\"%s\" not valid log page id.\n",
1009 				    optarg);
1010 				logpage_usage();
1011 			}
1012 			pageflag = true;
1013 			break;
1014 		case 'x':
1015 			hexflag = true;
1016 			break;
1017 		case 'v':
1018 			if (strcmp(optarg, "help") == 0)
1019 				logpage_help();
1020 			vendor = optarg;
1021 			break;
1022 		}
1023 	}
1024 
1025 	if (!pageflag) {
1026 		printf("Missing page_id (-p).\n");
1027 		logpage_usage();
1028 	}
1029 
1030 	/* Check that a controller and/or namespace was specified. */
1031 	if (optind >= argc)
1032 		logpage_usage();
1033 
1034 	if (strstr(argv[optind], NVME_NS_PREFIX) != NULL) {
1035 		ns_specified = true;
1036 		parse_ns_str(argv[optind], cname, &nsid);
1037 		open_dev(cname, &fd, 1, 1);
1038 	} else {
1039 		ns_specified = false;
1040 		nsid = 0xffffffff;
1041 		open_dev(argv[optind], &fd, 1, 1);
1042 	}
1043 
1044 	read_controller_data(fd, &cdata);
1045 
1046 	/*
1047 	 * The log page attributes indicate whether or not the controller
1048 	 * supports the SMART/Health information log page on a per
1049 	 * namespace basis.
1050 	 */
1051 	if (ns_specified) {
1052 		if (log_page != NVME_LOG_HEALTH_INFORMATION)
1053 			errx(1, "log page %d valid only at controller level",
1054 			    log_page);
1055 		if (!(cdata.lpa & NVME_ID_CTRLR_LPA_NS_SMART))
1056 			errx(1,
1057 			    "controller does not support per namespace "
1058 			    "smart/health information");
1059 	}
1060 
1061 	print_fn = print_log_hex;
1062 	size = DEFAULT_SIZE;
1063 	if (binflag)
1064 		print_fn = print_bin;
1065 	if (!binflag && !hexflag) {
1066 		/*
1067 		 * See if there is a pretty print function for the specified log
1068 		 * page.  If one isn't found, we just revert to the default
1069 		 * (print_hex). If there was a vendor specified bt the user, and
1070 		 * the page is vendor specific, don't match the print function
1071 		 * unless the vendors match.
1072 		 */
1073 		for (f = logfuncs; f->log_page > 0; f++) {
1074 			if (f->vendor != NULL && vendor != NULL &&
1075 			    strcmp(f->vendor, vendor) != 0)
1076 				continue;
1077 			if (log_page != f->log_page)
1078 				continue;
1079 			print_fn = f->print_fn;
1080 			size = f->size;
1081 			break;
1082 		}
1083 	}
1084 
1085 	if (log_page == NVME_LOG_ERROR) {
1086 		size = sizeof(struct nvme_error_information_entry);
1087 		size *= (cdata.elpe + 1);
1088 	}
1089 
1090 	/* Read the log page */
1091 	buf = get_log_buffer(size);
1092 	read_logpage(fd, log_page, nsid, buf, size);
1093 	print_fn(&cdata, buf, size);
1094 
1095 	close(fd);
1096 	exit(0);
1097 }
1098