1 // SPDX-License-Identifier: GPL-2.0 2 /* 3 * Hwmon client for disk and solid state drives with temperature sensors 4 * Copyright (C) 2019 Zodiac Inflight Innovations 5 * 6 * With input from: 7 * Hwmon client for S.M.A.R.T. hard disk drives with temperature sensors. 8 * (C) 2018 Linus Walleij 9 * 10 * hwmon: Driver for SCSI/ATA temperature sensors 11 * by Constantin Baranov <const@mimas.ru>, submitted September 2009 12 * 13 * This drive supports reporting the temperatire of SATA drives. It can be 14 * easily extended to report the temperature of SCSI drives. 15 * 16 * The primary means to read drive temperatures and temperature limits 17 * for ATA drives is the SCT Command Transport feature set as specified in 18 * ATA8-ACS. 19 * It can be used to read the current drive temperature, temperature limits, 20 * and historic minimum and maximum temperatures. The SCT Command Transport 21 * feature set is documented in "AT Attachment 8 - ATA/ATAPI Command Set 22 * (ATA8-ACS)". 23 * 24 * If the SCT Command Transport feature set is not available, drive temperatures 25 * may be readable through SMART attributes. Since SMART attributes are not well 26 * defined, this method is only used as fallback mechanism. 27 * 28 * There are three SMART attributes which may report drive temperatures. 29 * Those are defined as follows (from 30 * http://www.cropel.com/library/smart-attribute-list.aspx). 31 * 32 * 190 Temperature Temperature, monitored by a sensor somewhere inside 33 * the drive. Raw value typicaly holds the actual 34 * temperature (hexadecimal) in its rightmost two digits. 35 * 36 * 194 Temperature Temperature, monitored by a sensor somewhere inside 37 * the drive. Raw value typicaly holds the actual 38 * temperature (hexadecimal) in its rightmost two digits. 39 * 40 * 231 Temperature Temperature, monitored by a sensor somewhere inside 41 * the drive. Raw value typicaly holds the actual 42 * temperature (hexadecimal) in its rightmost two digits. 43 * 44 * Wikipedia defines attributes a bit differently. 45 * 46 * 190 Temperature Value is equal to (100-temp. °C), allowing manufacturer 47 * Difference or to set a minimum threshold which corresponds to a 48 * Airflow maximum temperature. This also follows the convention of 49 * Temperature 100 being a best-case value and lower values being 50 * undesirable. However, some older drives may instead 51 * report raw Temperature (identical to 0xC2) or 52 * Temperature minus 50 here. 53 * 194 Temperature or Indicates the device temperature, if the appropriate 54 * Temperature sensor is fitted. Lowest byte of the raw value contains 55 * Celsius the exact temperature value (Celsius degrees). 56 * 231 Life Left Indicates the approximate SSD life left, in terms of 57 * (SSDs) or program/erase cycles or available reserved blocks. 58 * Temperature A normalized value of 100 represents a new drive, with 59 * a threshold value at 10 indicating a need for 60 * replacement. A value of 0 may mean that the drive is 61 * operating in read-only mode to allow data recovery. 62 * Previously (pre-2010) occasionally used for Drive 63 * Temperature (more typically reported at 0xC2). 64 * 65 * Common denominator is that the first raw byte reports the temperature 66 * in degrees C on almost all drives. Some drives may report a fractional 67 * temperature in the second raw byte. 68 * 69 * Known exceptions (from libatasmart): 70 * - SAMSUNG SV0412H and SAMSUNG SV1204H) report the temperature in 10th 71 * degrees C in the first two raw bytes. 72 * - A few Maxtor drives report an unknown or bad value in attribute 194. 73 * - Certain Apple SSD drives report an unknown value in attribute 190. 74 * Only certain firmware versions are affected. 75 * 76 * Those exceptions affect older ATA drives and are currently ignored. 77 * Also, the second raw byte (possibly reporting the fractional temperature) 78 * is currently ignored. 79 * 80 * Many drives also report temperature limits in additional SMART data raw 81 * bytes. The format of those is not well defined and varies widely. 82 * The driver does not currently attempt to report those limits. 83 * 84 * According to data in smartmontools, attribute 231 is rarely used to report 85 * drive temperatures. At the same time, several drives report SSD life left 86 * in attribute 231, but do not support temperature sensors. For this reason, 87 * attribute 231 is currently ignored. 88 * 89 * Following above definitions, temperatures are reported as follows. 90 * If SCT Command Transport is supported, it is used to read the 91 * temperature and, if available, temperature limits. 92 * - Otherwise, if SMART attribute 194 is supported, it is used to read 93 * the temperature. 94 * - Otherwise, if SMART attribute 190 is supported, it is used to read 95 * the temperature. 96 */ 97 98 #include <linux/ata.h> 99 #include <linux/bits.h> 100 #include <linux/device.h> 101 #include <linux/hwmon.h> 102 #include <linux/kernel.h> 103 #include <linux/list.h> 104 #include <linux/module.h> 105 #include <linux/mutex.h> 106 #include <scsi/scsi_cmnd.h> 107 #include <scsi/scsi_device.h> 108 #include <scsi/scsi_driver.h> 109 #include <scsi/scsi_proto.h> 110 111 struct drivetemp_data { 112 struct list_head list; /* list of instantiated devices */ 113 struct mutex lock; /* protect data buffer accesses */ 114 struct scsi_device *sdev; /* SCSI device */ 115 struct device *dev; /* instantiating device */ 116 struct device *hwdev; /* hardware monitoring device */ 117 u8 smartdata[ATA_SECT_SIZE]; /* local buffer */ 118 int (*get_temp)(struct drivetemp_data *st, u32 attr, long *val); 119 bool have_temp_lowest; /* lowest temp in SCT status */ 120 bool have_temp_highest; /* highest temp in SCT status */ 121 bool have_temp_min; /* have min temp */ 122 bool have_temp_max; /* have max temp */ 123 bool have_temp_lcrit; /* have lower critical limit */ 124 bool have_temp_crit; /* have critical limit */ 125 int temp_min; /* min temp */ 126 int temp_max; /* max temp */ 127 int temp_lcrit; /* lower critical limit */ 128 int temp_crit; /* critical limit */ 129 }; 130 131 static LIST_HEAD(drivetemp_devlist); 132 133 #define ATA_MAX_SMART_ATTRS 30 134 #define SMART_TEMP_PROP_190 190 135 #define SMART_TEMP_PROP_194 194 136 137 #define SCT_STATUS_REQ_ADDR 0xe0 138 #define SCT_STATUS_VERSION_LOW 0 /* log byte offsets */ 139 #define SCT_STATUS_VERSION_HIGH 1 140 #define SCT_STATUS_TEMP 200 141 #define SCT_STATUS_TEMP_LOWEST 201 142 #define SCT_STATUS_TEMP_HIGHEST 202 143 #define SCT_READ_LOG_ADDR 0xe1 144 #define SMART_READ_LOG 0xd5 145 #define SMART_WRITE_LOG 0xd6 146 147 #define INVALID_TEMP 0x80 148 149 #define temp_is_valid(temp) ((temp) != INVALID_TEMP) 150 #define temp_from_sct(temp) (((s8)(temp)) * 1000) 151 152 static inline bool ata_id_smart_supported(u16 *id) 153 { 154 return id[ATA_ID_COMMAND_SET_1] & BIT(0); 155 } 156 157 static inline bool ata_id_smart_enabled(u16 *id) 158 { 159 return id[ATA_ID_CFS_ENABLE_1] & BIT(0); 160 } 161 162 static int drivetemp_scsi_command(struct drivetemp_data *st, 163 u8 ata_command, u8 feature, 164 u8 lba_low, u8 lba_mid, u8 lba_high) 165 { 166 u8 scsi_cmd[MAX_COMMAND_SIZE]; 167 int data_dir; 168 169 memset(scsi_cmd, 0, sizeof(scsi_cmd)); 170 scsi_cmd[0] = ATA_16; 171 if (ata_command == ATA_CMD_SMART && feature == SMART_WRITE_LOG) { 172 scsi_cmd[1] = (5 << 1); /* PIO Data-out */ 173 /* 174 * No off.line or cc, write to dev, block count in sector count 175 * field. 176 */ 177 scsi_cmd[2] = 0x06; 178 data_dir = DMA_TO_DEVICE; 179 } else { 180 scsi_cmd[1] = (4 << 1); /* PIO Data-in */ 181 /* 182 * No off.line or cc, read from dev, block count in sector count 183 * field. 184 */ 185 scsi_cmd[2] = 0x0e; 186 data_dir = DMA_FROM_DEVICE; 187 } 188 scsi_cmd[4] = feature; 189 scsi_cmd[6] = 1; /* 1 sector */ 190 scsi_cmd[8] = lba_low; 191 scsi_cmd[10] = lba_mid; 192 scsi_cmd[12] = lba_high; 193 scsi_cmd[14] = ata_command; 194 195 return scsi_execute_req(st->sdev, scsi_cmd, data_dir, 196 st->smartdata, ATA_SECT_SIZE, NULL, HZ, 5, 197 NULL); 198 } 199 200 static int drivetemp_ata_command(struct drivetemp_data *st, u8 feature, 201 u8 select) 202 { 203 return drivetemp_scsi_command(st, ATA_CMD_SMART, feature, select, 204 ATA_SMART_LBAM_PASS, ATA_SMART_LBAH_PASS); 205 } 206 207 static int drivetemp_get_smarttemp(struct drivetemp_data *st, u32 attr, 208 long *temp) 209 { 210 u8 *buf = st->smartdata; 211 bool have_temp = false; 212 u8 temp_raw; 213 u8 csum; 214 int err; 215 int i; 216 217 err = drivetemp_ata_command(st, ATA_SMART_READ_VALUES, 0); 218 if (err) 219 return err; 220 221 /* Checksum the read value table */ 222 csum = 0; 223 for (i = 0; i < ATA_SECT_SIZE; i++) 224 csum += buf[i]; 225 if (csum) { 226 dev_dbg(&st->sdev->sdev_gendev, 227 "checksum error reading SMART values\n"); 228 return -EIO; 229 } 230 231 for (i = 0; i < ATA_MAX_SMART_ATTRS; i++) { 232 u8 *attr = buf + i * 12; 233 int id = attr[2]; 234 235 if (!id) 236 continue; 237 238 if (id == SMART_TEMP_PROP_190) { 239 temp_raw = attr[7]; 240 have_temp = true; 241 } 242 if (id == SMART_TEMP_PROP_194) { 243 temp_raw = attr[7]; 244 have_temp = true; 245 break; 246 } 247 } 248 249 if (have_temp) { 250 *temp = temp_raw * 1000; 251 return 0; 252 } 253 254 return -ENXIO; 255 } 256 257 static int drivetemp_get_scttemp(struct drivetemp_data *st, u32 attr, long *val) 258 { 259 u8 *buf = st->smartdata; 260 int err; 261 262 err = drivetemp_ata_command(st, SMART_READ_LOG, SCT_STATUS_REQ_ADDR); 263 if (err) 264 return err; 265 switch (attr) { 266 case hwmon_temp_input: 267 *val = temp_from_sct(buf[SCT_STATUS_TEMP]); 268 break; 269 case hwmon_temp_lowest: 270 *val = temp_from_sct(buf[SCT_STATUS_TEMP_LOWEST]); 271 break; 272 case hwmon_temp_highest: 273 *val = temp_from_sct(buf[SCT_STATUS_TEMP_HIGHEST]); 274 break; 275 default: 276 err = -EINVAL; 277 break; 278 } 279 return err; 280 } 281 282 static int drivetemp_identify_sata(struct drivetemp_data *st) 283 { 284 struct scsi_device *sdev = st->sdev; 285 u8 *buf = st->smartdata; 286 struct scsi_vpd *vpd; 287 bool is_ata, is_sata; 288 bool have_sct_data_table; 289 bool have_sct_temp; 290 bool have_smart; 291 bool have_sct; 292 u16 *ata_id; 293 u16 version; 294 long temp; 295 int err; 296 297 /* SCSI-ATA Translation present? */ 298 rcu_read_lock(); 299 vpd = rcu_dereference(sdev->vpd_pg89); 300 301 /* 302 * Verify that ATA IDENTIFY DEVICE data is included in ATA Information 303 * VPD and that the drive implements the SATA protocol. 304 */ 305 if (!vpd || vpd->len < 572 || vpd->data[56] != ATA_CMD_ID_ATA || 306 vpd->data[36] != 0x34) { 307 rcu_read_unlock(); 308 return -ENODEV; 309 } 310 ata_id = (u16 *)&vpd->data[60]; 311 is_ata = ata_id_is_ata(ata_id); 312 is_sata = ata_id_is_sata(ata_id); 313 have_sct = ata_id_sct_supported(ata_id); 314 have_sct_data_table = ata_id_sct_data_tables(ata_id); 315 have_smart = ata_id_smart_supported(ata_id) && 316 ata_id_smart_enabled(ata_id); 317 318 rcu_read_unlock(); 319 320 /* bail out if this is not a SATA device */ 321 if (!is_ata || !is_sata) 322 return -ENODEV; 323 if (!have_sct) 324 goto skip_sct; 325 326 err = drivetemp_ata_command(st, SMART_READ_LOG, SCT_STATUS_REQ_ADDR); 327 if (err) 328 goto skip_sct; 329 330 version = (buf[SCT_STATUS_VERSION_HIGH] << 8) | 331 buf[SCT_STATUS_VERSION_LOW]; 332 if (version != 2 && version != 3) 333 goto skip_sct; 334 335 have_sct_temp = temp_is_valid(buf[SCT_STATUS_TEMP]); 336 if (!have_sct_temp) 337 goto skip_sct; 338 339 st->have_temp_lowest = temp_is_valid(buf[SCT_STATUS_TEMP_LOWEST]); 340 st->have_temp_highest = temp_is_valid(buf[SCT_STATUS_TEMP_HIGHEST]); 341 342 if (!have_sct_data_table) 343 goto skip_sct; 344 345 /* Request and read temperature history table */ 346 memset(buf, '\0', sizeof(st->smartdata)); 347 buf[0] = 5; /* data table command */ 348 buf[2] = 1; /* read table */ 349 buf[4] = 2; /* temperature history table */ 350 351 err = drivetemp_ata_command(st, SMART_WRITE_LOG, SCT_STATUS_REQ_ADDR); 352 if (err) 353 goto skip_sct_data; 354 355 err = drivetemp_ata_command(st, SMART_READ_LOG, SCT_READ_LOG_ADDR); 356 if (err) 357 goto skip_sct_data; 358 359 /* 360 * Temperature limits per AT Attachment 8 - 361 * ATA/ATAPI Command Set (ATA8-ACS) 362 */ 363 st->have_temp_max = temp_is_valid(buf[6]); 364 st->have_temp_crit = temp_is_valid(buf[7]); 365 st->have_temp_min = temp_is_valid(buf[8]); 366 st->have_temp_lcrit = temp_is_valid(buf[9]); 367 368 st->temp_max = temp_from_sct(buf[6]); 369 st->temp_crit = temp_from_sct(buf[7]); 370 st->temp_min = temp_from_sct(buf[8]); 371 st->temp_lcrit = temp_from_sct(buf[9]); 372 373 skip_sct_data: 374 if (have_sct_temp) { 375 st->get_temp = drivetemp_get_scttemp; 376 return 0; 377 } 378 skip_sct: 379 if (!have_smart) 380 return -ENODEV; 381 st->get_temp = drivetemp_get_smarttemp; 382 return drivetemp_get_smarttemp(st, hwmon_temp_input, &temp); 383 } 384 385 static int drivetemp_identify(struct drivetemp_data *st) 386 { 387 struct scsi_device *sdev = st->sdev; 388 389 /* Bail out immediately if there is no inquiry data */ 390 if (!sdev->inquiry || sdev->inquiry_len < 16) 391 return -ENODEV; 392 393 /* Disk device? */ 394 if (sdev->type != TYPE_DISK && sdev->type != TYPE_ZBC) 395 return -ENODEV; 396 397 return drivetemp_identify_sata(st); 398 } 399 400 static int drivetemp_read(struct device *dev, enum hwmon_sensor_types type, 401 u32 attr, int channel, long *val) 402 { 403 struct drivetemp_data *st = dev_get_drvdata(dev); 404 int err = 0; 405 406 if (type != hwmon_temp) 407 return -EINVAL; 408 409 switch (attr) { 410 case hwmon_temp_input: 411 case hwmon_temp_lowest: 412 case hwmon_temp_highest: 413 mutex_lock(&st->lock); 414 err = st->get_temp(st, attr, val); 415 mutex_unlock(&st->lock); 416 break; 417 case hwmon_temp_lcrit: 418 *val = st->temp_lcrit; 419 break; 420 case hwmon_temp_min: 421 *val = st->temp_min; 422 break; 423 case hwmon_temp_max: 424 *val = st->temp_max; 425 break; 426 case hwmon_temp_crit: 427 *val = st->temp_crit; 428 break; 429 default: 430 err = -EINVAL; 431 break; 432 } 433 return err; 434 } 435 436 static umode_t drivetemp_is_visible(const void *data, 437 enum hwmon_sensor_types type, 438 u32 attr, int channel) 439 { 440 const struct drivetemp_data *st = data; 441 442 switch (type) { 443 case hwmon_temp: 444 switch (attr) { 445 case hwmon_temp_input: 446 return 0444; 447 case hwmon_temp_lowest: 448 if (st->have_temp_lowest) 449 return 0444; 450 break; 451 case hwmon_temp_highest: 452 if (st->have_temp_highest) 453 return 0444; 454 break; 455 case hwmon_temp_min: 456 if (st->have_temp_min) 457 return 0444; 458 break; 459 case hwmon_temp_max: 460 if (st->have_temp_max) 461 return 0444; 462 break; 463 case hwmon_temp_lcrit: 464 if (st->have_temp_lcrit) 465 return 0444; 466 break; 467 case hwmon_temp_crit: 468 if (st->have_temp_crit) 469 return 0444; 470 break; 471 default: 472 break; 473 } 474 break; 475 default: 476 break; 477 } 478 return 0; 479 } 480 481 static const struct hwmon_channel_info *drivetemp_info[] = { 482 HWMON_CHANNEL_INFO(chip, 483 HWMON_C_REGISTER_TZ), 484 HWMON_CHANNEL_INFO(temp, HWMON_T_INPUT | 485 HWMON_T_LOWEST | HWMON_T_HIGHEST | 486 HWMON_T_MIN | HWMON_T_MAX | 487 HWMON_T_LCRIT | HWMON_T_CRIT), 488 NULL 489 }; 490 491 static const struct hwmon_ops drivetemp_ops = { 492 .is_visible = drivetemp_is_visible, 493 .read = drivetemp_read, 494 }; 495 496 static const struct hwmon_chip_info drivetemp_chip_info = { 497 .ops = &drivetemp_ops, 498 .info = drivetemp_info, 499 }; 500 501 /* 502 * The device argument points to sdev->sdev_dev. Its parent is 503 * sdev->sdev_gendev, which we can use to get the scsi_device pointer. 504 */ 505 static int drivetemp_add(struct device *dev, struct class_interface *intf) 506 { 507 struct scsi_device *sdev = to_scsi_device(dev->parent); 508 struct drivetemp_data *st; 509 int err; 510 511 st = kzalloc(sizeof(*st), GFP_KERNEL); 512 if (!st) 513 return -ENOMEM; 514 515 st->sdev = sdev; 516 st->dev = dev; 517 mutex_init(&st->lock); 518 519 if (drivetemp_identify(st)) { 520 err = -ENODEV; 521 goto abort; 522 } 523 524 st->hwdev = hwmon_device_register_with_info(dev->parent, "drivetemp", 525 st, &drivetemp_chip_info, 526 NULL); 527 if (IS_ERR(st->hwdev)) { 528 err = PTR_ERR(st->hwdev); 529 goto abort; 530 } 531 532 list_add(&st->list, &drivetemp_devlist); 533 return 0; 534 535 abort: 536 kfree(st); 537 return err; 538 } 539 540 static void drivetemp_remove(struct device *dev, struct class_interface *intf) 541 { 542 struct drivetemp_data *st, *tmp; 543 544 list_for_each_entry_safe(st, tmp, &drivetemp_devlist, list) { 545 if (st->dev == dev) { 546 list_del(&st->list); 547 hwmon_device_unregister(st->hwdev); 548 kfree(st); 549 break; 550 } 551 } 552 } 553 554 static struct class_interface drivetemp_interface = { 555 .add_dev = drivetemp_add, 556 .remove_dev = drivetemp_remove, 557 }; 558 559 static int __init drivetemp_init(void) 560 { 561 return scsi_register_interface(&drivetemp_interface); 562 } 563 564 static void __exit drivetemp_exit(void) 565 { 566 scsi_unregister_interface(&drivetemp_interface); 567 } 568 569 module_init(drivetemp_init); 570 module_exit(drivetemp_exit); 571 572 MODULE_AUTHOR("Guenter Roeck <linus@roeck-us.net>"); 573 MODULE_DESCRIPTION("Hard drive temperature monitor"); 574 MODULE_LICENSE("GPL"); 575