2 * Home page of code is: http://www.smartmontools.org
4 * Copyright (C) 2002-11 Bruce Allen
5 * Copyright (C) 2008-17 Christian Franke
6 * Copyright (C) 2000 Michael Cornwell <cornwell@acm.org>
7 * Copyright (C) 2008 Oliver Bock <brevilo@users.sourceforge.net>
9 * This program is free software; you can redistribute it and/or modify
10 * it under the terms of the GNU General Public License as published by
11 * the Free Software Foundation; either version 2, or (at your option)
14 * You should have received a copy of the GNU General Public License
15 * (for example COPYING); If not, see <http://www.gnu.org/licenses/>.
17 * This code was originally developed as a Senior Thesis by Michael Cornwell
18 * at the Concurrent Systems Laboratory (now part of the Storage Systems
19 * Research Center), Jack Baskin School of Engineering, University of
20 * California, Santa Cruz. http://ssrc.soe.ucsc.edu/
27 // unconditionally included files
29 #include <sys/types.h>
30 #include <sys/stat.h> // umask
45 #include <algorithm> // std::replace()
47 // conditionally included files
57 #pragma warning(disable:4761) // "conversion supplied"
58 typedef unsigned short mode_t
;
61 #include <io.h> // umask()
62 #include <process.h> // getpid()
66 #include <io.h> // setmode()
73 // locally included files
75 #include "dev_interface.h"
76 #include "knowndrives.h"
81 // This is for solaris, where signal() resets the handler to SIG_DFL
82 // after the first signal is caught.
84 #define SIGNALFN sigset
86 #define SIGNALFN signal
90 // fork()/signal()/initd simulation for native Windows
91 #include "daemon_win32.h" // daemon_main/detach/signal()
93 #define SIGNALFN daemon_signal
94 #define strsignal daemon_strsignal
95 #define sleep daemon_sleep
96 // SIGQUIT does not exist, CONTROL-Break signals SIGBREAK.
97 #define SIGQUIT SIGBREAK
98 #define SIGQUIT_KEYNAME "CONTROL-Break"
100 #define SIGQUIT_KEYNAME "CONTROL-\\"
103 const char * smartd_cpp_cvsid
= "$Id: smartd.cpp 4556 2017-10-15 17:31:30Z chrfranke $"
106 using namespace smartmontools
;
109 #define EXIT_BADCMD 1 // command line did not parse
110 #define EXIT_BADCONF 2 // syntax error in config file
111 #define EXIT_STARTUP 3 // problem forking daemon
112 #define EXIT_PID 4 // problem creating pid file
113 #define EXIT_NOCONF 5 // config file does not exist
114 #define EXIT_READCONF 6 // config file exists but cannot be read
116 #define EXIT_NOMEM 8 // out of memory
117 #define EXIT_BADCODE 10 // internal error - should NEVER happen
119 #define EXIT_BADDEV 16 // we can't monitor this device
120 #define EXIT_NODEV 17 // no devices to monitor
122 #define EXIT_SIGNAL 254 // abort on signal
125 // command-line: 1=debug mode, 2=print presets
126 static unsigned char debugmode
= 0;
128 // command-line: how long to sleep between checks
129 #define CHECKTIME 1800
130 static int checktime
=CHECKTIME
;
132 // command-line: name of PID file (empty for no pid file)
133 static std::string pid_file
;
135 // command-line: path prefix of persistent state file, empty if no persistence.
136 static std::string state_path_prefix
137 #ifdef SMARTMONTOOLS_SAVESTATES
138 = SMARTMONTOOLS_SAVESTATES
142 // command-line: path prefix of attribute log file, empty if no logs.
143 static std::string attrlog_path_prefix
144 #ifdef SMARTMONTOOLS_ATTRIBUTELOG
145 = SMARTMONTOOLS_ATTRIBUTELOG
149 // configuration file name
150 static const char * configfile
;
151 // configuration file "name" if read from stdin
152 static const char * const configfile_stdin
= "<stdin>";
153 // path of alternate configuration file
154 static std::string configfile_alt
;
156 // warning script file
157 static std::string warning_script
;
159 // command-line: when should we exit?
161 QUIT_NODEV
, QUIT_NODEVSTARTUP
, QUIT_NEVER
, QUIT_ONECHECK
,
162 QUIT_SHOWTESTS
, QUIT_ERRORS
164 static quit_t quit
= QUIT_NODEV
;
166 // command-line; this is the default syslog(3) log facility to use.
167 static int facility
=LOG_DAEMON
;
170 // command-line: fork into background?
171 static bool do_fork
=true;
174 #ifdef HAVE_LIBCAP_NG
175 // command-line: enable capabilities?
176 static bool enable_capabilities
= false;
179 // TODO: This smartctl only variable is also used in os_win32.cpp
180 unsigned char failuretest_permissive
= 0;
182 // set to one if we catch a USR1 (check devices now)
183 static volatile int caughtsigUSR1
=0;
186 // set to one if we catch a USR2 (toggle debug mode)
187 static volatile int caughtsigUSR2
=0;
190 // set to one if we catch a HUP (reload config file). In debug mode,
191 // set to two, if we catch INT (also reload config file).
192 static volatile int caughtsigHUP
=0;
194 // set to signal value if we catch INT, QUIT, or TERM
195 static volatile int caughtsigEXIT
=0;
197 // This function prints either to stdout or to the syslog as needed.
198 static void PrintOut(int priority
, const char *fmt
, ...)
199 __attribute_format_printf(2, 3);
201 // Attribute monitoring flags.
202 // See monitor_attr_flags below.
204 MONITOR_IGN_FAILUSE
= 0x01,
205 MONITOR_IGNORE
= 0x02,
206 MONITOR_RAW_PRINT
= 0x04,
208 MONITOR_AS_CRIT
= 0x10,
209 MONITOR_RAW_AS_CRIT
= 0x20,
212 // Array of flags for each attribute.
213 class attribute_flags
217 { memset(m_flags
, 0, sizeof(m_flags
)); }
219 bool is_set(int id
, unsigned char flag
) const
220 { return (0 < id
&& id
< (int)sizeof(m_flags
) && (m_flags
[id
] & flag
)); }
222 void set(int id
, unsigned char flags
)
224 if (0 < id
&& id
< (int)sizeof(m_flags
))
225 m_flags
[id
] |= flags
;
229 unsigned char m_flags
[256];
233 /// Configuration data for a device. Read from smartd.conf.
234 /// Supports copy & assignment and is compatible with STL containers.
237 int lineno
; // Line number of entry in file
238 std::string name
; // Device name (with optional extra info)
239 std::string dev_name
; // Device name (plain, for SMARTD_DEVICE variable)
240 std::string dev_type
; // Device type argument from -d directive, empty if none
241 std::string dev_idinfo
; // Device identify info for warning emails
242 std::string state_file
; // Path of the persistent state file, empty if none
243 std::string attrlog_file
; // Path of the persistent attrlog file, empty if none
244 bool ignore
; // Ignore this entry
245 bool id_is_unique
; // True if dev_idinfo is unique (includes S/N or WWN)
246 bool smartcheck
; // Check SMART status
247 bool usagefailed
; // Check for failed Usage Attributes
248 bool prefail
; // Track changes in Prefail Attributes
249 bool usage
; // Track changes in Usage Attributes
250 bool selftest
; // Monitor number of selftest errors
251 bool errorlog
; // Monitor number of ATA errors
252 bool xerrorlog
; // Monitor number of ATA errors (Extended Comprehensive error log)
253 bool offlinests
; // Monitor changes in offline data collection status
254 bool offlinests_ns
; // Disable auto standby if in progress
255 bool selfteststs
; // Monitor changes in self-test execution status
256 bool selfteststs_ns
; // Disable auto standby if in progress
257 bool permissive
; // Ignore failed SMART commands
258 char autosave
; // 1=disable, 2=enable Autosave Attributes
259 char autoofflinetest
; // 1=disable, 2=enable Auto Offline Test
260 firmwarebug_defs firmwarebugs
; // -F directives from drivedb or smartd.conf
261 bool ignorepresets
; // Ignore database of -v options
262 bool showpresets
; // Show database entry for this device
263 bool removable
; // Device may disappear (not be present)
264 char powermode
; // skip check, if disk in idle or standby mode
265 bool powerquiet
; // skip powermode 'skipping checks' message
266 int powerskipmax
; // how many times can be check skipped
267 unsigned char tempdiff
; // Track Temperature changes >= this limit
268 unsigned char tempinfo
, tempcrit
; // Track Temperatures >= these limits as LOG_INFO, LOG_CRIT+mail
269 regular_expression test_regex
; // Regex for scheduled testing
271 // Configuration of email warning messages
272 std::string emailcmdline
; // script to execute, empty if no messages
273 std::string emailaddress
; // email address, or empty
274 unsigned char emailfreq
; // Emails once (1) daily (2) diminishing (3)
275 bool emailtest
; // Send test email?
278 int dev_rpm
; // rotation rate, 0 = unknown, 1 = SSD, >1 = HDD
279 int set_aam
; // disable(-1), enable(1..255->0..254) Automatic Acoustic Management
280 int set_apm
; // disable(-1), enable(2..255->1..254) Advanced Power Management
281 int set_lookahead
; // disable(-1), enable(1) read look-ahead
282 int set_standby
; // set(1..255->0..254) standby timer
283 bool set_security_freeze
; // Freeze ATA security
284 int set_wcache
; // disable(-1), enable(1) write cache
285 int set_dsn
; // disable(0x2), enable(0x1) DSN
287 bool sct_erc_set
; // set SCT ERC to:
288 unsigned short sct_erc_readtime
; // ERC read time (deciseconds)
289 unsigned short sct_erc_writetime
; // ERC write time (deciseconds)
291 unsigned char curr_pending_id
; // ID of current pending sector count, 0 if none
292 unsigned char offl_pending_id
; // ID of offline uncorrectable sector count, 0 if none
293 bool curr_pending_incr
, offl_pending_incr
; // True if current/offline pending values increase
294 bool curr_pending_set
, offl_pending_set
; // True if '-C', '-U' set in smartd.conf
296 attribute_flags monitor_attr_flags
; // MONITOR_* flags for each attribute
298 ata_vendor_attr_defs attribute_defs
; // -v options
303 dev_config::dev_config()
314 offlinests(false), offlinests_ns(false),
315 selfteststs(false), selfteststs_ns(false),
319 ignorepresets(false),
326 tempinfo(0), tempcrit(0),
330 set_aam(0), set_apm(0),
333 set_security_freeze(false),
334 set_wcache(0), set_dsn(0),
336 sct_erc_readtime(0), sct_erc_writetime(0),
337 curr_pending_id(0), offl_pending_id(0),
338 curr_pending_incr(false), offl_pending_incr(false),
339 curr_pending_set(false), offl_pending_set(false)
344 // Number of allowed mail message types
345 static const int SMARTD_NMAIL
= 13;
346 // Type for '-M test' mails (state not persistent)
347 static const int MAILTYPE_TEST
= 0;
348 // TODO: Add const or enum for all mail types.
351 int logged
;// number of times an email has been sent
352 time_t firstsent
;// time first email was sent, as defined by time(2)
353 time_t lastsent
; // time last email was sent, as defined by time(2)
356 : logged(0), firstsent(0), lastsent(0) { }
359 /// Persistent state data for a device.
360 struct persistent_dev_state
362 unsigned char tempmin
, tempmax
; // Min/Max Temperatures
364 unsigned char selflogcount
; // total number of self-test errors
365 unsigned short selfloghour
; // lifetime hours of last self-test error
367 time_t scheduled_test_next_check
; // Time of next check for scheduled self-tests
369 uint64_t selective_test_last_start
; // Start LBA of last scheduled selective self-test
370 uint64_t selective_test_last_end
; // End LBA of last scheduled selective self-test
372 mailinfo maillog
[SMARTD_NMAIL
]; // log info on when mail sent
375 int ataerrorcount
; // Total number of ATA errors
377 // Persistent part of ata_smart_values:
378 struct ata_attribute
{
381 unsigned char worst
; // Byte needed for 'raw64' attribute only.
385 ata_attribute() : id(0), val(0), worst(0), raw(0), resvd(0) { }
387 ata_attribute ata_attributes
[NUMBER_ATA_SMART_ATTRIBUTES
];
391 struct scsi_error_counter_t
{
392 struct scsiErrorCounter errCounter
;
394 scsi_error_counter_t() : found(0)
395 { memset(&errCounter
, 0, sizeof(errCounter
)); }
397 scsi_error_counter_t scsi_error_counters
[3];
399 struct scsi_nonmedium_error_t
{
400 struct scsiNonMediumError nme
;
402 scsi_nonmedium_error_t() : found(0)
403 { memset(&nme
, 0, sizeof(nme
)); }
405 scsi_nonmedium_error_t scsi_nonmedium_error
;
408 uint64_t nvme_err_log_entries
;
410 persistent_dev_state();
413 persistent_dev_state::persistent_dev_state()
414 : tempmin(0), tempmax(0),
417 scheduled_test_next_check(0),
418 selective_test_last_start(0),
419 selective_test_last_end(0),
421 nvme_err_log_entries(0)
425 /// Non-persistent state data for a device.
426 struct temp_dev_state
428 bool must_write
; // true if persistent part should be written
430 bool not_cap_offline
; // true == not capable of offline testing
431 bool not_cap_conveyance
;
434 bool not_cap_selective
;
436 unsigned char temperature
; // last recorded Temperature (in Celsius)
437 time_t tempmin_delay
; // time where Min Temperature tracking will start
439 bool removed
; // true if open() failed for removable device
441 bool powermodefail
; // true if power mode check failed
442 int powerskipcnt
; // Number of checks skipped due to idle or standby mode
443 int lastpowermodeskipped
; // the last power mode that was skipped
446 unsigned char SmartPageSupported
; // has log sense IE page (0x2f)
447 unsigned char TempPageSupported
; // has log sense temperature page (0xd)
448 unsigned char ReadECounterPageSupported
;
449 unsigned char WriteECounterPageSupported
;
450 unsigned char VerifyECounterPageSupported
;
451 unsigned char NonMediumErrorPageSupported
;
452 unsigned char SuppressReport
; // minimize nuisance reports
453 unsigned char modese_len
; // mode sense/select cmd len: 0 (don't
456 uint64_t num_sectors
; // Number of sectors
457 ata_smart_values smartval
; // SMART data
458 ata_smart_thresholds_pvt smartthres
; // SMART thresholds
459 bool offline_started
; // true if offline data collection was started
460 bool selftest_started
; // true if self-test was started
465 temp_dev_state::temp_dev_state()
467 not_cap_offline(false),
468 not_cap_conveyance(false),
469 not_cap_short(false),
471 not_cap_selective(false),
475 powermodefail(false),
477 lastpowermodeskipped(0),
478 SmartPageSupported(false),
479 TempPageSupported(false),
480 ReadECounterPageSupported(false),
481 WriteECounterPageSupported(false),
482 VerifyECounterPageSupported(false),
483 NonMediumErrorPageSupported(false),
484 SuppressReport(false),
487 offline_started(false),
488 selftest_started(false)
490 memset(&smartval
, 0, sizeof(smartval
));
491 memset(&smartthres
, 0, sizeof(smartthres
));
494 /// Runtime state data for a device.
496 : public persistent_dev_state
,
497 public temp_dev_state
499 void update_persistent_state();
500 void update_temp_state();
503 /// Container for configuration info for each device.
504 typedef std::vector
<dev_config
> dev_config_vector
;
506 /// Container for state info for each device.
507 typedef std::vector
<dev_state
> dev_state_vector
;
509 // Copy ATA attributes to persistent state.
510 void dev_state::update_persistent_state()
512 for (int i
= 0; i
< NUMBER_ATA_SMART_ATTRIBUTES
; i
++) {
513 const ata_smart_attribute
& ta
= smartval
.vendor_attributes
[i
];
514 ata_attribute
& pa
= ata_attributes
[i
];
517 pa
.val
= pa
.worst
= 0; pa
.raw
= 0;
525 | ((uint64_t)ta
.raw
[3] << 24)
526 | ((uint64_t)ta
.raw
[4] << 32)
527 | ((uint64_t)ta
.raw
[5] << 40);
528 pa
.resvd
= ta
.reserv
;
532 // Copy ATA from persistent to temp state.
533 void dev_state::update_temp_state()
535 for (int i
= 0; i
< NUMBER_ATA_SMART_ATTRIBUTES
; i
++) {
536 const ata_attribute
& pa
= ata_attributes
[i
];
537 ata_smart_attribute
& ta
= smartval
.vendor_attributes
[i
];
540 ta
.current
= ta
.worst
= 0;
541 memset(ta
.raw
, 0, sizeof(ta
.raw
));
546 ta
.raw
[0] = (unsigned char) pa
.raw
;
547 ta
.raw
[1] = (unsigned char)(pa
.raw
>> 8);
548 ta
.raw
[2] = (unsigned char)(pa
.raw
>> 16);
549 ta
.raw
[3] = (unsigned char)(pa
.raw
>> 24);
550 ta
.raw
[4] = (unsigned char)(pa
.raw
>> 32);
551 ta
.raw
[5] = (unsigned char)(pa
.raw
>> 40);
552 ta
.reserv
= pa
.resvd
;
556 // Parse a line from a state file.
557 static bool parse_dev_state_line(const char * line
, persistent_dev_state
& state
)
559 static const regular_expression
regex(
561 "((temperature-min)" // (1 (2)
562 "|(temperature-max)" // (3)
563 "|(self-test-errors)" // (4)
564 "|(self-test-last-err-hour)" // (5)
565 "|(scheduled-test-next-check)" // (6)
566 "|(selective-test-last-start)" // (7)
567 "|(selective-test-last-end)" // (8)
568 "|(ata-error-count)" // (9)
569 "|(mail\\.([0-9]+)\\." // (10 (11)
570 "((count)" // (12 (13)
571 "|(first-sent-time)" // (14)
572 "|(last-sent-time)" // (15)
575 "|(ata-smart-attribute\\.([0-9]+)\\." // (16 (17)
583 "|(nvme-err-log-entries)" // (24)
585 " *= *([0-9]+)[ \n]*$", // (25)
589 const int nmatch
= 1+25;
590 regmatch_t match
[nmatch
];
591 if (!regex
.execute(line
, nmatch
, match
))
593 if (match
[nmatch
-1].rm_so
< 0)
596 uint64_t val
= strtoull(line
+ match
[nmatch
-1].rm_so
, (char **)0, 10);
599 if (match
[++m
].rm_so
>= 0)
600 state
.tempmin
= (unsigned char)val
;
601 else if (match
[++m
].rm_so
>= 0)
602 state
.tempmax
= (unsigned char)val
;
603 else if (match
[++m
].rm_so
>= 0)
604 state
.selflogcount
= (unsigned char)val
;
605 else if (match
[++m
].rm_so
>= 0)
606 state
.selfloghour
= (unsigned short)val
;
607 else if (match
[++m
].rm_so
>= 0)
608 state
.scheduled_test_next_check
= (time_t)val
;
609 else if (match
[++m
].rm_so
>= 0)
610 state
.selective_test_last_start
= val
;
611 else if (match
[++m
].rm_so
>= 0)
612 state
.selective_test_last_end
= val
;
613 else if (match
[++m
].rm_so
>= 0)
614 state
.ataerrorcount
= (int)val
;
615 else if (match
[m
+=2].rm_so
>= 0) {
616 int i
= atoi(line
+match
[m
].rm_so
);
617 if (!(0 <= i
&& i
< SMARTD_NMAIL
))
619 if (i
== MAILTYPE_TEST
) // Don't suppress test mails
621 if (match
[m
+=2].rm_so
>= 0)
622 state
.maillog
[i
].logged
= (int)val
;
623 else if (match
[++m
].rm_so
>= 0)
624 state
.maillog
[i
].firstsent
= (time_t)val
;
625 else if (match
[++m
].rm_so
>= 0)
626 state
.maillog
[i
].lastsent
= (time_t)val
;
630 else if (match
[m
+=5+1].rm_so
>= 0) {
631 int i
= atoi(line
+match
[m
].rm_so
);
632 if (!(0 <= i
&& i
< NUMBER_ATA_SMART_ATTRIBUTES
))
634 if (match
[m
+=2].rm_so
>= 0)
635 state
.ata_attributes
[i
].id
= (unsigned char)val
;
636 else if (match
[++m
].rm_so
>= 0)
637 state
.ata_attributes
[i
].val
= (unsigned char)val
;
638 else if (match
[++m
].rm_so
>= 0)
639 state
.ata_attributes
[i
].worst
= (unsigned char)val
;
640 else if (match
[++m
].rm_so
>= 0)
641 state
.ata_attributes
[i
].raw
= val
;
642 else if (match
[++m
].rm_so
>= 0)
643 state
.ata_attributes
[i
].resvd
= (unsigned char)val
;
647 else if (match
[m
+7].rm_so
>= 0)
648 state
.nvme_err_log_entries
= val
;
654 // Read a state file.
655 static bool read_dev_state(const char * path
, persistent_dev_state
& state
)
657 stdio_file
f(path
, "r");
660 pout("Cannot read state file \"%s\"\n", path
);
664 setmode(fileno(f
), O_TEXT
); // Allow files with \r\n
667 persistent_dev_state new_state
;
668 int good
= 0, bad
= 0;
670 while (fgets(line
, sizeof(line
), f
)) {
671 const char * s
= line
+ strspn(line
, " \t");
672 if (!*s
|| *s
== '#')
674 if (!parse_dev_state_line(line
, new_state
))
682 pout("%s: format error\n", path
);
685 pout("%s: %d invalid line(s) ignored\n", path
, bad
);
688 // This sets the values missing in the file to 0.
693 static void write_dev_state_line(FILE * f
, const char * name
, uint64_t val
)
696 fprintf(f
, "%s = %" PRIu64
"\n", name
, val
);
699 static void write_dev_state_line(FILE * f
, const char * name1
, int id
, const char * name2
, uint64_t val
)
702 fprintf(f
, "%s.%d.%s = %" PRIu64
"\n", name1
, id
, name2
, val
);
705 // Write a state file
706 static bool write_dev_state(const char * path
, const persistent_dev_state
& state
)
708 // Rename old "file" to "file~"
709 std::string pathbak
= path
; pathbak
+= '~';
710 unlink(pathbak
.c_str());
711 rename(path
, pathbak
.c_str());
713 stdio_file
f(path
, "w");
715 pout("Cannot create state file \"%s\"\n", path
);
719 fprintf(f
, "# smartd state file\n");
720 write_dev_state_line(f
, "temperature-min", state
.tempmin
);
721 write_dev_state_line(f
, "temperature-max", state
.tempmax
);
722 write_dev_state_line(f
, "self-test-errors", state
.selflogcount
);
723 write_dev_state_line(f
, "self-test-last-err-hour", state
.selfloghour
);
724 write_dev_state_line(f
, "scheduled-test-next-check", state
.scheduled_test_next_check
);
725 write_dev_state_line(f
, "selective-test-last-start", state
.selective_test_last_start
);
726 write_dev_state_line(f
, "selective-test-last-end", state
.selective_test_last_end
);
729 for (i
= 0; i
< SMARTD_NMAIL
; i
++) {
730 if (i
== MAILTYPE_TEST
) // Don't suppress test mails
732 const mailinfo
& mi
= state
.maillog
[i
];
735 write_dev_state_line(f
, "mail", i
, "count", mi
.logged
);
736 write_dev_state_line(f
, "mail", i
, "first-sent-time", mi
.firstsent
);
737 write_dev_state_line(f
, "mail", i
, "last-sent-time", mi
.lastsent
);
741 write_dev_state_line(f
, "ata-error-count", state
.ataerrorcount
);
743 for (i
= 0; i
< NUMBER_ATA_SMART_ATTRIBUTES
; i
++) {
744 const persistent_dev_state::ata_attribute
& pa
= state
.ata_attributes
[i
];
747 write_dev_state_line(f
, "ata-smart-attribute", i
, "id", pa
.id
);
748 write_dev_state_line(f
, "ata-smart-attribute", i
, "val", pa
.val
);
749 write_dev_state_line(f
, "ata-smart-attribute", i
, "worst", pa
.worst
);
750 write_dev_state_line(f
, "ata-smart-attribute", i
, "raw", pa
.raw
);
751 write_dev_state_line(f
, "ata-smart-attribute", i
, "resvd", pa
.resvd
);
755 write_dev_state_line(f
, "nvme-err-log-entries", state
.nvme_err_log_entries
);
760 // Write to the attrlog file
761 static bool write_dev_attrlog(const char * path
, const dev_state
& state
)
763 stdio_file
f(path
, "a");
765 pout("Cannot create attribute log file \"%s\"\n", path
);
770 time_t now
= time(0);
771 struct tm
* tms
= gmtime(&now
);
772 fprintf(f
, "%d-%02d-%02d %02d:%02d:%02d;",
773 1900+tms
->tm_year
, 1+tms
->tm_mon
, tms
->tm_mday
,
774 tms
->tm_hour
, tms
->tm_min
, tms
->tm_sec
);
776 for (int i
= 0; i
< NUMBER_ATA_SMART_ATTRIBUTES
; i
++) {
777 const persistent_dev_state::ata_attribute
& pa
= state
.ata_attributes
[i
];
780 fprintf(f
, "\t%d;%d;%" PRIu64
";", pa
.id
, pa
.val
, pa
.raw
);
783 const struct scsiErrorCounter
* ecp
;
784 const char * pageNames
[3] = {"read", "write", "verify"};
785 for (int k
= 0; k
< 3; ++k
) {
786 if ( !state
.scsi_error_counters
[k
].found
) continue;
787 ecp
= &state
.scsi_error_counters
[k
].errCounter
;
788 fprintf(f
, "\t%s-corr-by-ecc-fast;%" PRIu64
";"
789 "\t%s-corr-by-ecc-delayed;%" PRIu64
";"
790 "\t%s-corr-by-retry;%" PRIu64
";"
791 "\t%s-total-err-corrected;%" PRIu64
";"
792 "\t%s-corr-algorithm-invocations;%" PRIu64
";"
793 "\t%s-gb-processed;%.3f;"
794 "\t%s-total-unc-errors;%" PRIu64
";",
795 pageNames
[k
], ecp
->counter
[0],
796 pageNames
[k
], ecp
->counter
[1],
797 pageNames
[k
], ecp
->counter
[2],
798 pageNames
[k
], ecp
->counter
[3],
799 pageNames
[k
], ecp
->counter
[4],
800 pageNames
[k
], (ecp
->counter
[5] / 1000000000.0),
801 pageNames
[k
], ecp
->counter
[6]);
803 if(state
.scsi_nonmedium_error
.found
&& state
.scsi_nonmedium_error
.nme
.gotPC0
) {
804 fprintf(f
, "\tnon-medium-errors;%" PRIu64
";", state
.scsi_nonmedium_error
.nme
.counterPC0
);
806 // write SCSI current temperature if it is monitored
807 if (state
.temperature
)
808 fprintf(f
, "\ttemperature;%d;", state
.temperature
);
814 // Write all state files. If write_always is false, don't write
815 // unless must_write is set.
816 static void write_all_dev_states(const dev_config_vector
& configs
,
817 dev_state_vector
& states
,
818 bool write_always
= true)
820 for (unsigned i
= 0; i
< states
.size(); i
++) {
821 const dev_config
& cfg
= configs
.at(i
);
822 if (cfg
.state_file
.empty())
824 dev_state
& state
= states
[i
];
825 if (!write_always
&& !state
.must_write
)
827 if (!write_dev_state(cfg
.state_file
.c_str(), state
))
829 state
.must_write
= false;
830 if (write_always
|| debugmode
)
831 PrintOut(LOG_INFO
, "Device: %s, state written to %s\n",
832 cfg
.name
.c_str(), cfg
.state_file
.c_str());
836 // Write to all attrlog files
837 static void write_all_dev_attrlogs(const dev_config_vector
& configs
,
838 dev_state_vector
& states
)
840 for (unsigned i
= 0; i
< states
.size(); i
++) {
841 const dev_config
& cfg
= configs
.at(i
);
842 if (cfg
.attrlog_file
.empty())
844 dev_state
& state
= states
[i
];
845 write_dev_attrlog(cfg
.attrlog_file
.c_str(), state
);
849 // remove the PID file
850 static void RemovePidFile()
852 if (!pid_file
.empty()) {
853 if (unlink(pid_file
.c_str()))
854 PrintOut(LOG_CRIT
,"Can't unlink PID file %s (%s).\n",
855 pid_file
.c_str(), strerror(errno
));
861 extern "C" { // signal handlers require C-linkage
863 // Note if we catch a SIGUSR1
864 static void USR1handler(int sig
)
872 // Note if we catch a SIGUSR2
873 static void USR2handler(int sig
)
881 // Note if we catch a HUP (or INT in debug mode)
882 static void HUPhandler(int sig
)
891 // signal handler for TERM, QUIT, and INT (if not in debug mode)
892 static void sighandler(int sig
)
901 // Cleanup, print Goodbye message and remove pidfile
902 static int Goodbye(int status
)
904 // delete PID file, if one was created
907 // and this should be the final output from smartd before it exits
908 PrintOut(status
?LOG_CRIT
:LOG_INFO
, "smartd is exiting (exit status %d)\n", status
);
913 // a replacement for setenv() which is not available on all platforms.
914 // Note that the string passed to putenv must not be freed or made
915 // invalid, since a pointer to it is kept by putenv(). This means that
916 // it must either be a static buffer or allocated off the heap. The
917 // string can be freed if the environment variable is redefined via
918 // another call to putenv(). There is no portable way to unset a variable
919 // with putenv(). So we manage the buffer in a static object.
920 // Using setenv() if available is not considered because some
921 // implementations may produce memory leaks.
927 : m_buf((char *)0) { }
929 void set(const char * name
, const char * value
);
934 env_buffer(const env_buffer
&);
935 void operator=(const env_buffer
&);
938 void env_buffer::set(const char * name
, const char * value
)
940 int size
= strlen(name
) + 1 + strlen(value
) + 1;
941 char * newbuf
= new char[size
];
942 snprintf(newbuf
, size
, "%s=%s", name
, value
);
945 throw std::runtime_error("putenv() failed");
947 // This assumes that the same NAME is passed on each call
954 static void MailWarning(const dev_config
& cfg
, dev_state
& state
, int which
, const char *fmt
, ...)
955 __attribute_format_printf(4, 5);
957 // If either address or executable path is non-null then send and log
958 // a warning email, or execute executable
959 static void MailWarning(const dev_config
& cfg
, dev_state
& state
, int which
, const char *fmt
, ...)
961 static const char * const whichfail
[] = {
967 "FailedHealthCheck", // 5
968 "FailedReadSmartData", // 6
969 "FailedReadSmartErrorLog", // 7
970 "FailedReadSmartSelfTestLog", // 8
971 "FailedOpenDevice", // 9
972 "CurrentPendingSector", // 10
973 "OfflineUncorrectableSector", // 11
977 // See if user wants us to send mail
978 if (cfg
.emailaddress
.empty() && cfg
.emailcmdline
.empty())
981 std::string address
= cfg
.emailaddress
;
982 const char * executable
= cfg
.emailcmdline
.c_str();
984 // which type of mail are we sending?
985 mailinfo
* mail
=(state
.maillog
)+which
;
988 if (cfg
.emailfreq
<1 || cfg
.emailfreq
>3) {
989 PrintOut(LOG_CRIT
,"internal error in MailWarning(): cfg.mailwarn->emailfreq=%d\n",cfg
.emailfreq
);
992 if (which
<0 || which
>=SMARTD_NMAIL
|| sizeof(whichfail
)!=SMARTD_NMAIL
*sizeof(char *)) {
993 PrintOut(LOG_CRIT
,"Contact " PACKAGE_BUGREPORT
"; internal error in MailWarning(): which=%d, size=%d\n",
994 which
, (int)sizeof(whichfail
));
998 // Return if a single warning mail has been sent.
999 if ((cfg
.emailfreq
==1) && mail
->logged
)
1002 // Return if this is an email test and one has already been sent.
1003 if (which
== 0 && mail
->logged
)
1006 // To decide if to send mail, we need to know what time it is.
1007 time_t epoch
= time(0);
1009 // Return if less than one day has gone by
1010 const int day
= 24*3600;
1011 if (cfg
.emailfreq
==2 && mail
->logged
&& epoch
<(mail
->lastsent
+day
))
1014 // Return if less than 2^(logged-1) days have gone by
1015 if (cfg
.emailfreq
==3 && mail
->logged
) {
1016 int days
= 0x01 << (mail
->logged
- 1);
1018 if (epoch
<(mail
->lastsent
+days
))
1022 #ifdef HAVE_LIBCAP_NG
1023 if (enable_capabilities
) {
1024 PrintOut(LOG_ERR
, "Sending a mail was supressed. "
1025 "Mails can't be send when capabilites are enabled\n");
1030 // record the time of this mail message, and the first mail message
1032 mail
->firstsent
=epoch
;
1033 mail
->lastsent
=epoch
;
1035 // print warning string into message
1039 vsnprintf(message
, sizeof(message
), fmt
, ap
);
1042 // replace commas by spaces to separate recipients
1043 std::replace(address
.begin(), address
.end(), ',', ' ');
1045 // Export information in environment variables that will be useful
1047 static env_buffer env
[12];
1048 env
[0].set("SMARTD_MAILER", executable
);
1049 env
[1].set("SMARTD_MESSAGE", message
);
1050 char dates
[DATEANDEPOCHLEN
];
1051 snprintf(dates
, sizeof(dates
), "%d", mail
->logged
);
1052 env
[2].set("SMARTD_PREVCNT", dates
);
1053 dateandtimezoneepoch(dates
, mail
->firstsent
);
1054 env
[3].set("SMARTD_TFIRST", dates
);
1055 snprintf(dates
, DATEANDEPOCHLEN
,"%d", (int)mail
->firstsent
);
1056 env
[4].set("SMARTD_TFIRSTEPOCH", dates
);
1057 env
[5].set("SMARTD_FAILTYPE", whichfail
[which
]);
1058 env
[6].set("SMARTD_ADDRESS", address
.c_str());
1059 env
[7].set("SMARTD_DEVICESTRING", cfg
.name
.c_str());
1061 // Allow 'smartctl ... -d $SMARTD_DEVICETYPE $SMARTD_DEVICE'
1062 env
[8].set("SMARTD_DEVICETYPE",
1063 (!cfg
.dev_type
.empty() ? cfg
.dev_type
.c_str() : "auto"));
1064 env
[9].set("SMARTD_DEVICE", cfg
.dev_name
.c_str());
1066 env
[10].set("SMARTD_DEVICEINFO", cfg
.dev_idinfo
.c_str());
1068 if (which
) switch (cfg
.emailfreq
) {
1069 case 2: dates
[0] = '1'; dates
[1] = 0; break;
1070 case 3: snprintf(dates
, sizeof(dates
), "%d", (0x01)<<mail
->logged
);
1072 env
[11].set("SMARTD_NEXTDAYS", dates
);
1074 // now construct a command to send this as EMAIL
1076 executable
= "<mail>";
1077 const char * newadd
= (!address
.empty()? address
.c_str() : "<nomailer>");
1078 const char * newwarn
= (which
? "Warning via" : "Test of");
1082 snprintf(command
, sizeof(command
), "%s 2>&1", warning_script
.c_str());
1084 // tell SYSLOG what we are about to do...
1085 PrintOut(LOG_INFO
,"%s %s to %s ...\n",
1086 which
?"Sending warning via":"Executing test of", executable
, newadd
);
1088 // issue the command to send mail or to run the user's executable
1091 if (!(pfp
=popen(command
, "r")))
1092 // failed to popen() mail process
1093 PrintOut(LOG_CRIT
,"%s %s to %s: failed (fork or pipe failed, or no memory) %s\n",
1094 newwarn
, executable
, newadd
, errno
?strerror(errno
):"");
1098 char buffer
[EBUFLEN
];
1100 // if unexpected output on stdout/stderr, null terminate, print, and flush
1101 if ((len
=fread(buffer
, 1, EBUFLEN
, pfp
))) {
1103 int newlen
= len
<EBUFLEN
? len
: EBUFLEN
-1;
1104 buffer
[newlen
]='\0';
1105 PrintOut(LOG_CRIT
,"%s %s to %s produced unexpected output (%s%d bytes) to STDOUT/STDERR: \n%s\n",
1106 newwarn
, executable
, newadd
, len
!=newlen
?"here truncated to ":"", newlen
, buffer
);
1108 // flush pipe if needed
1109 while (fread(buffer
, 1, EBUFLEN
, pfp
) && count
<EBUFLEN
)
1112 // tell user that pipe was flushed, or that something is really wrong
1113 if (count
&& count
<EBUFLEN
)
1114 PrintOut(LOG_CRIT
,"%s %s to %s: flushed remaining STDOUT/STDERR\n",
1115 newwarn
, executable
, newadd
);
1117 PrintOut(LOG_CRIT
,"%s %s to %s: more than 1 MB STDOUT/STDERR flushed, breaking pipe\n",
1118 newwarn
, executable
, newadd
);
1121 // if something went wrong with mail process, print warning
1123 if (-1==(status
=pclose(pfp
)))
1124 PrintOut(LOG_CRIT
,"%s %s to %s: pclose(3) failed %s\n", newwarn
, executable
, newadd
,
1125 errno
?strerror(errno
):"");
1127 // mail process apparently succeeded. Check and report exit status
1128 if (WIFEXITED(status
)) {
1129 // exited 'normally' (but perhaps with nonzero status)
1130 int status8
= WEXITSTATUS(status
);
1132 PrintOut(LOG_CRIT
,"%s %s to %s: failed (32-bit/8-bit exit status: %d/%d) perhaps caught signal %d [%s]\n",
1133 newwarn
, executable
, newadd
, status
, status8
, status8
-128, strsignal(status8
-128));
1135 PrintOut(LOG_CRIT
,"%s %s to %s: failed (32-bit/8-bit exit status: %d/%d)\n",
1136 newwarn
, executable
, newadd
, status
, status8
);
1138 PrintOut(LOG_INFO
,"%s %s to %s: successful\n", newwarn
, executable
, newadd
);
1141 if (WIFSIGNALED(status
))
1142 PrintOut(LOG_INFO
,"%s %s to %s: exited because of uncaught signal %d [%s]\n",
1143 newwarn
, executable
, newadd
, WTERMSIG(status
), strsignal(WTERMSIG(status
)));
1145 // this branch is probably not possible. If subprocess is
1146 // stopped then pclose() should not return.
1147 if (WIFSTOPPED(status
))
1148 PrintOut(LOG_CRIT
,"%s %s to %s: process STOPPED because it caught signal %d [%s]\n",
1149 newwarn
, executable
, newadd
, WSTOPSIG(status
), strsignal(WSTOPSIG(status
)));
1157 snprintf(command
, sizeof(command
), "cmd /c \"%s\"", warning_script
.c_str());
1159 char stdoutbuf
[800]; // < buffer in syslog_win32::vsyslog()
1162 PrintOut(LOG_INFO
,"%s %s to %s ...\n",
1163 (which
?"Sending warning via":"Executing test of"), executable
, newadd
);
1164 rc
= daemon_spawn(command
, "", 0, stdoutbuf
, sizeof(stdoutbuf
));
1165 if (rc
>= 0 && stdoutbuf
[0])
1166 PrintOut(LOG_CRIT
,"%s %s to %s produced unexpected output (%d bytes) to STDOUT/STDERR:\n%s\n",
1167 newwarn
, executable
, newadd
, (int)strlen(stdoutbuf
), stdoutbuf
);
1169 PrintOut(LOG_CRIT
,"%s %s to %s: failed, exit status %d\n",
1170 newwarn
, executable
, newadd
, rc
);
1172 PrintOut(LOG_INFO
,"%s %s to %s: successful\n", newwarn
, executable
, newadd
);
1177 // increment mail sent counter
1181 static void reset_warning_mail(const dev_config
& cfg
, dev_state
& state
, int which
, const char *fmt
, ...)
1182 __attribute_format_printf(4, 5);
1184 static void reset_warning_mail(const dev_config
& cfg
, dev_state
& state
, int which
, const char *fmt
, ...)
1186 if (!(0 <= which
&& which
< SMARTD_NMAIL
))
1189 // Return if no mail sent yet
1190 mailinfo
& mi
= state
.maillog
[which
];
1194 // Format & print message
1198 vsnprintf(msg
, sizeof(msg
), fmt
, ap
);
1201 PrintOut(LOG_INFO
, "Device: %s, %s, warning condition reset after %d email%s\n", cfg
.name
.c_str(),
1202 msg
, mi
.logged
, (mi
.logged
==1 ? "" : "s"));
1204 // Clear mail counter and timestamps
1206 state
.must_write
= true;
1211 // Output multiple lines via separate syslog(3) calls.
1212 __attribute_format_printf(2, 0)
1213 static void vsyslog_lines(int priority
, const char * fmt
, va_list ap
)
1215 char buf
[512+EBUFLEN
]; // enough space for exec cmd output in MailWarning()
1216 vsnprintf(buf
, sizeof(buf
), fmt
, ap
);
1218 for (char * p
= buf
, * q
; p
&& *p
; p
= q
) {
1219 if ((q
= strchr(p
, '\n')))
1222 syslog(priority
, "%s\n", p
);
1227 // os_win32/syslog_win32.cpp supports multiple lines.
1228 #define vsyslog_lines vsyslog
1231 // Printing function for watching ataprint commands, or losing them
1232 // [From GLIBC Manual: Since the prototype doesn't specify types for
1233 // optional arguments, in a call to a variadic function the default
1234 // argument promotions are performed on the optional argument
1235 // values. This means the objects of type char or short int (whether
1236 // signed or not) are promoted to either int or unsigned int, as
1238 void pout(const char *fmt
, ...){
1241 // get the correct time in syslog()
1242 FixGlibcTimeZoneBug();
1243 // initialize variable argument list
1245 // in debugmode==1 mode we will print the output from the ataprint.o functions!
1246 if (debugmode
&& debugmode
!= 2) {
1249 if (facility
== LOG_LOCAL1
) // logging to stdout
1252 vfprintf(f
, fmt
, ap
);
1255 // in debugmode==2 mode we print output from knowndrives.o functions
1256 else if (debugmode
==2 || ata_debugmode
|| scsi_debugmode
) {
1257 openlog("smartd", LOG_PID
, facility
);
1258 vsyslog_lines(LOG_INFO
, fmt
, ap
);
1265 // This function prints either to stdout or to the syslog as needed.
1266 static void PrintOut(int priority
, const char *fmt
, ...){
1269 // get the correct time in syslog()
1270 FixGlibcTimeZoneBug();
1271 // initialize variable argument list
1276 if (facility
== LOG_LOCAL1
) // logging to stdout
1279 vfprintf(f
, fmt
, ap
);
1283 openlog("smartd", LOG_PID
, facility
);
1284 vsyslog_lines(priority
, fmt
, ap
);
1291 // Used to warn users about invalid checksums. Called from atacmds.cpp.
1292 void checksumwarning(const char * string
)
1294 pout("Warning! %s error: invalid SMART checksum.\n", string
);
1299 // Wait for the pid file to show up, this makes sure a calling program knows
1300 // that the daemon is really up and running and has a pid to kill it
1301 static bool WaitForPidFile()
1303 int waited
, max_wait
= 10;
1304 struct stat stat_buf
;
1306 if (pid_file
.empty() || debugmode
)
1309 for(waited
= 0; waited
< max_wait
; ++waited
) {
1310 if (!stat(pid_file
.c_str(), &stat_buf
)) {
1320 // Forks new process, closes ALL file descriptors, redirects stdin,
1321 // stdout, and stderr. Not quite daemon(). See
1322 // http://www.linuxjournal.com/article/2335
1323 // for a good description of why we do things this way.
1324 static void DaemonInit()
1328 // flush all buffered streams. Else we might get two copies of open
1329 // streams since both parent and child get copies of the buffers.
1334 if ((pid
=fork()) < 0) {
1336 PrintOut(LOG_CRIT
,"smartd unable to fork daemon process!\n");
1340 // we are the parent process, wait for pid file, then exit cleanly
1341 if(!WaitForPidFile()) {
1342 PrintOut(LOG_CRIT
,"PID file %s didn't show up!\n", pid_file
.c_str());
1348 // from here on, we are the child process.
1351 // Fork one more time to avoid any possibility of having terminals
1352 if ((pid
=fork()) < 0) {
1354 PrintOut(LOG_CRIT
,"smartd unable to fork daemon process!\n");
1358 // we are the parent process -- exit cleanly
1361 // Now we are the child's child...
1364 // close any open file descriptors
1365 for (int i
= getdtablesize(); --i
>= 0; )
1368 // redirect any IO attempts to /dev/null and change to root directory
1369 int fd
= open("/dev/null", O_RDWR
);
1370 if (!(fd
== 0 && dup(fd
) == 1 && dup(fd
) == 2 && !chdir("/"))) {
1371 PrintOut(LOG_CRIT
, "smartd unable to redirect to /dev/null or to chdir to root!\n");
1377 PrintOut(LOG_INFO
, "smartd has fork()ed into background mode. New PID=%d.\n", (int)getpid());
1381 // No fork() on native Win32
1382 // Detach this process from console
1384 if (daemon_detach("smartd")) {
1385 PrintOut(LOG_CRIT
,"smartd unable to detach from console!\n");
1388 // stdin/out/err now closed if not redirected
1394 // create a PID file containing the current process id
1395 static void WritePidFile()
1397 if (!pid_file
.empty()) {
1398 pid_t pid
= getpid();
1401 old_umask
= umask(0077); // rwx------
1403 // Cygwin: smartd service runs on system account, ensure PID file can be read by admins
1404 old_umask
= umask(0033); // rwxr--r--
1407 stdio_file
f(pid_file
.c_str(), "w");
1409 if (!(f
&& fprintf(f
, "%d\n", (int)pid
) > 0 && f
.close())) {
1410 PrintOut(LOG_CRIT
, "unable to write PID file %s - exiting.\n", pid_file
.c_str());
1413 PrintOut(LOG_INFO
, "file %s written containing PID %d\n", pid_file
.c_str(), (int)pid
);
1417 // Prints header identifying version of code and home
1418 static void PrintHead()
1420 PrintOut(LOG_INFO
, "%s\n", format_version_info("smartd").c_str());
1423 // prints help info for configuration file Directives
1424 static void Directives()
1427 "Configuration file (%s) Directives (after device name):\n"
1428 " -d TYPE Set the device type: auto, ignore, removable,\n"
1430 " -T TYPE Set the tolerance to one of: normal, permissive\n"
1431 " -o VAL Enable/disable automatic offline tests (on/off)\n"
1432 " -S VAL Enable/disable attribute autosave (on/off)\n"
1433 " -n MODE No check if: never, sleep[,N][,q], standby[,N][,q], idle[,N][,q]\n"
1434 " -H Monitor SMART Health Status, report if failed\n"
1435 " -s REG Do Self-Test at time(s) given by regular expression REG\n"
1436 " -l TYPE Monitor SMART log or self-test status:\n"
1437 " error, selftest, xerror, offlinests[,ns], selfteststs[,ns]\n"
1438 " -l scterc,R,W Set SCT Error Recovery Control\n"
1439 " -e Change device setting: aam,[N|off], apm,[N|off], dsn,[on|off],\n"
1440 " lookahead,[on|off], security-freeze, standby,[N|off], wcache,[on|off]\n"
1441 " -f Monitor 'Usage' Attributes, report failures\n"
1442 " -m ADD Send email warning to address ADD\n"
1443 " -M TYPE Modify email warning behavior (see man page)\n"
1444 " -p Report changes in 'Prefailure' Attributes\n"
1445 " -u Report changes in 'Usage' Attributes\n"
1446 " -t Equivalent to -p and -u Directives\n"
1447 " -r ID Also report Raw values of Attribute ID with -p, -u or -t\n"
1448 " -R ID Track changes in Attribute ID Raw value with -p, -u or -t\n"
1449 " -i ID Ignore Attribute ID for -f Directive\n"
1450 " -I ID Ignore Attribute ID for -p, -u or -t Directive\n"
1451 " -C ID[+] Monitor [increases of] Current Pending Sectors in Attribute ID\n"
1452 " -U ID[+] Monitor [increases of] Offline Uncorrectable Sectors in Attribute ID\n"
1453 " -W D,I,C Monitor Temperature D)ifference, I)nformal limit, C)ritical limit\n"
1454 " -v N,ST Modifies labeling of Attribute N (see man page) \n"
1455 " -P TYPE Drive-specific presets: use, ignore, show, showall\n"
1456 " -a Default: -H -f -t -l error -l selftest -l selfteststs -C 197 -U 198\n"
1457 " -F TYPE Use firmware bug workaround:\n"
1459 " # Comment: text after a hash sign is ignored\n"
1460 " \\ Line continuation character\n"
1461 "Attribute ID is a decimal integer 1 <= ID <= 255\n"
1462 "Use ID = 0 to turn off -C and/or -U Directives\n"
1463 "Example: /dev/sda -a\n",
1465 smi()->get_valid_dev_types_str().c_str(),
1466 get_valid_firmwarebug_args());
1469 /* Returns a pointer to a static string containing a formatted list of the valid
1470 arguments to the option opt or NULL on failure. */
1471 static const char *GetValidArgList(char opt
)
1476 return "<PATH_PREFIX>";
1478 return "[+]<FILE_NAME>";
1480 return "<FILE_NAME>, -";
1482 return "daemon, local0, local1, local2, local3, local4, local5, local6, local7";
1484 return "nodev, errors, nodevstartup, never, onecheck, showtests";
1486 return "ioctl[,N], ataioctl[,N], scsiioctl[,N], nvmeioctl[,N]";
1489 return "<FILE_NAME>";
1491 return "<INTEGER_SECONDS>";
1497 /* prints help information for command syntax */
1500 PrintOut(LOG_INFO
,"Usage: smartd [options]\n\n");
1501 PrintOut(LOG_INFO
," -A PREFIX, --attributelog=PREFIX\n");
1502 PrintOut(LOG_INFO
," Log ATA attribute information to {PREFIX}MODEL-SERIAL.ata.csv\n");
1503 #ifdef SMARTMONTOOLS_ATTRIBUTELOG
1504 PrintOut(LOG_INFO
," [default is " SMARTMONTOOLS_ATTRIBUTELOG
"MODEL-SERIAL.ata.csv]\n");
1506 PrintOut(LOG_INFO
,"\n");
1507 PrintOut(LOG_INFO
," -B [+]FILE, --drivedb=[+]FILE\n");
1508 PrintOut(LOG_INFO
," Read and replace [add] drive database from FILE\n");
1509 PrintOut(LOG_INFO
," [default is +%s", get_drivedb_path_add());
1510 #ifdef SMARTMONTOOLS_DRIVEDBDIR
1511 PrintOut(LOG_INFO
,"\n");
1512 PrintOut(LOG_INFO
," and then %s", get_drivedb_path_default());
1514 PrintOut(LOG_INFO
,"]\n\n");
1515 PrintOut(LOG_INFO
," -c NAME|-, --configfile=NAME|-\n");
1516 PrintOut(LOG_INFO
," Read configuration file NAME or stdin\n");
1517 PrintOut(LOG_INFO
," [default is %s]\n\n", configfile
);
1518 #ifdef HAVE_LIBCAP_NG
1519 PrintOut(LOG_INFO
," -C, --capabilities\n");
1520 PrintOut(LOG_INFO
," Drop unneeded Linux process capabilities.\n"
1521 " Warning: Mail notification does not work when used.\n\n");
1523 PrintOut(LOG_INFO
," -d, --debug\n");
1524 PrintOut(LOG_INFO
," Start smartd in debug mode\n\n");
1525 PrintOut(LOG_INFO
," -D, --showdirectives\n");
1526 PrintOut(LOG_INFO
," Print the configuration file Directives and exit\n\n");
1527 PrintOut(LOG_INFO
," -h, --help, --usage\n");
1528 PrintOut(LOG_INFO
," Display this help and exit\n\n");
1529 PrintOut(LOG_INFO
," -i N, --interval=N\n");
1530 PrintOut(LOG_INFO
," Set interval between disk checks to N seconds, where N >= 10\n\n");
1531 PrintOut(LOG_INFO
," -l local[0-7], --logfacility=local[0-7]\n");
1533 PrintOut(LOG_INFO
," Use syslog facility local0 - local7 or daemon [default]\n\n");
1535 PrintOut(LOG_INFO
," Log to \"./smartd.log\", stdout, stderr [default is event log]\n\n");
1538 PrintOut(LOG_INFO
," -n, --no-fork\n");
1539 PrintOut(LOG_INFO
," Do not fork into background\n\n");
1541 PrintOut(LOG_INFO
," -p NAME, --pidfile=NAME\n");
1542 PrintOut(LOG_INFO
," Write PID file NAME\n\n");
1543 PrintOut(LOG_INFO
," -q WHEN, --quit=WHEN\n");
1544 PrintOut(LOG_INFO
," Quit on one of: %s\n\n", GetValidArgList('q'));
1545 PrintOut(LOG_INFO
," -r, --report=TYPE\n");
1546 PrintOut(LOG_INFO
," Report transactions for one of: %s\n\n", GetValidArgList('r'));
1547 PrintOut(LOG_INFO
," -s PREFIX, --savestates=PREFIX\n");
1548 PrintOut(LOG_INFO
," Save disk states to {PREFIX}MODEL-SERIAL.TYPE.state\n");
1549 #ifdef SMARTMONTOOLS_SAVESTATES
1550 PrintOut(LOG_INFO
," [default is " SMARTMONTOOLS_SAVESTATES
"MODEL-SERIAL.TYPE.state]\n");
1552 PrintOut(LOG_INFO
,"\n");
1553 PrintOut(LOG_INFO
," -w NAME, --warnexec=NAME\n");
1554 PrintOut(LOG_INFO
," Run executable NAME on warnings\n");
1556 PrintOut(LOG_INFO
," [default is " SMARTMONTOOLS_SMARTDSCRIPTDIR
"/smartd_warning.sh]\n\n");
1558 PrintOut(LOG_INFO
," [default is %s/smartd_warning.cmd]\n\n", get_exe_dir().c_str());
1561 PrintOut(LOG_INFO
," --service\n");
1562 PrintOut(LOG_INFO
," Running as windows service (see man page), install with:\n");
1563 PrintOut(LOG_INFO
," smartd install [options]\n");
1564 PrintOut(LOG_INFO
," Remove service with:\n");
1565 PrintOut(LOG_INFO
," smartd remove\n\n");
1567 PrintOut(LOG_INFO
," -V, --version, --license, --copyright\n");
1568 PrintOut(LOG_INFO
," Print License, Copyright, and version information\n");
1571 static int CloseDevice(smart_device
* device
, const char * name
)
1573 if (!device
->close()){
1574 PrintOut(LOG_INFO
,"Device: %s, %s, close() failed\n", name
, device
->get_errmsg());
1577 // device sucessfully closed
1581 // return true if a char is not allowed in a state file name
1582 static bool not_allowed_in_filename(char c
)
1584 return !( ('0' <= c
&& c
<= '9')
1585 || ('A' <= c
&& c
<= 'Z')
1586 || ('a' <= c
&& c
<= 'z'));
1589 // Read error count from Summary or Extended Comprehensive SMART error log
1590 // Return -1 on error
1591 static int read_ata_error_count(ata_device
* device
, const char * name
,
1592 firmwarebug_defs firmwarebugs
, bool extended
)
1595 ata_smart_errorlog log
;
1596 if (ataReadErrorLog(device
, &log
, firmwarebugs
)){
1597 PrintOut(LOG_INFO
,"Device: %s, Read Summary SMART Error Log failed\n",name
);
1600 return (log
.error_log_pointer
? log
.ata_error_count
: 0);
1603 ata_smart_exterrlog logx
;
1604 if (!ataReadExtErrorLog(device
, &logx
, 0, 1 /*first sector only*/, firmwarebugs
)) {
1605 PrintOut(LOG_INFO
,"Device: %s, Read Extended Comprehensive SMART Error Log failed\n",name
);
1608 // Some disks use the reserved byte as index, see ataprint.cpp.
1609 return (logx
.error_log_index
|| logx
.reserved1
? logx
.device_error_count
: 0);
1613 // returns <0 if problem. Otherwise, bottom 8 bits are the self test
1614 // error count, and top bits are the power-on hours of the last error.
1615 static int SelfTestErrorCount(ata_device
* device
, const char * name
,
1616 firmwarebug_defs firmwarebugs
)
1618 struct ata_smart_selftestlog log
;
1620 if (ataReadSelfTestLog(device
, &log
, firmwarebugs
)){
1621 PrintOut(LOG_INFO
,"Device: %s, Read SMART Self Test Log Failed\n",name
);
1625 // return current number of self-test errors
1626 return ataPrintSmartSelfTestlog(&log
, false, firmwarebugs
);
1629 #define SELFTEST_ERRORCOUNT(x) (x & 0xff)
1630 #define SELFTEST_ERRORHOURS(x) ((x >> 8) & 0xffff)
1632 // Check offline data collection status
1633 static inline bool is_offl_coll_in_progress(unsigned char status
)
1635 return ((status
& 0x7f) == 0x03);
1638 // Check self-test execution status
1639 static inline bool is_self_test_in_progress(unsigned char status
)
1641 return ((status
>> 4) == 0xf);
1644 // Log offline data collection status
1645 static void log_offline_data_coll_status(const char * name
, unsigned char status
)
1648 switch (status
& 0x7f) {
1649 case 0x00: msg
= "was never started"; break;
1650 case 0x02: msg
= "was completed without error"; break;
1651 case 0x03: msg
= "is in progress"; break;
1652 case 0x04: msg
= "was suspended by an interrupting command from host"; break;
1653 case 0x05: msg
= "was aborted by an interrupting command from host"; break;
1654 case 0x06: msg
= "was aborted by the device with a fatal error"; break;
1659 PrintOut(((status
& 0x7f) == 0x06 ? LOG_CRIT
: LOG_INFO
),
1660 "Device: %s, offline data collection %s%s\n", name
, msg
,
1661 ((status
& 0x80) ? " (auto:on)" : ""));
1663 PrintOut(LOG_INFO
, "Device: %s, unknown offline data collection status 0x%02x\n",
1667 // Log self-test execution status
1668 static void log_self_test_exec_status(const char * name
, unsigned char status
)
1671 switch (status
>> 4) {
1672 case 0x0: msg
= "completed without error"; break;
1673 case 0x1: msg
= "was aborted by the host"; break;
1674 case 0x2: msg
= "was interrupted by the host with a reset"; break;
1675 case 0x3: msg
= "could not complete due to a fatal or unknown error"; break;
1676 case 0x4: msg
= "completed with error (unknown test element)"; break;
1677 case 0x5: msg
= "completed with error (electrical test element)"; break;
1678 case 0x6: msg
= "completed with error (servo/seek test element)"; break;
1679 case 0x7: msg
= "completed with error (read test element)"; break;
1680 case 0x8: msg
= "completed with error (handling damage?)"; break;
1685 PrintOut(((status
>> 4) >= 0x4 ? LOG_CRIT
: LOG_INFO
),
1686 "Device: %s, previous self-test %s\n", name
, msg
);
1687 else if ((status
>> 4) == 0xf)
1688 PrintOut(LOG_INFO
, "Device: %s, self-test in progress, %u0%% remaining\n",
1689 name
, status
& 0x0f);
1691 PrintOut(LOG_INFO
, "Device: %s, unknown self-test status 0x%02x\n",
1695 // Check pending sector count id (-C, -U directives).
1696 static bool check_pending_id(const dev_config
& cfg
, const dev_state
& state
,
1697 unsigned char id
, const char * msg
)
1699 // Check attribute index
1700 int i
= ata_find_attr_index(id
, state
.smartval
);
1702 PrintOut(LOG_INFO
, "Device: %s, can't monitor %s count - no Attribute %d\n",
1703 cfg
.name
.c_str(), msg
, id
);
1708 uint64_t rawval
= ata_get_attr_raw_value(state
.smartval
.vendor_attributes
[i
],
1709 cfg
.attribute_defs
);
1710 if (rawval
>= (state
.num_sectors
? state
.num_sectors
: 0xffffffffULL
)) {
1711 PrintOut(LOG_INFO
, "Device: %s, ignoring %s count - bogus Attribute %d value %" PRIu64
" (0x%" PRIx64
")\n",
1712 cfg
.name
.c_str(), msg
, id
, rawval
, rawval
);
1719 // Called by ATA/SCSI/NVMeDeviceScan() after successful device check
1720 static void finish_device_scan(dev_config
& cfg
, dev_state
& state
)
1722 // Set cfg.emailfreq if user hasn't set it
1723 if ((!cfg
.emailaddress
.empty() || !cfg
.emailcmdline
.empty()) && !cfg
.emailfreq
) {
1724 // Avoid that emails are suppressed forever due to state persistence
1725 if (cfg
.state_file
.empty())
1726 cfg
.emailfreq
= 1; // '-M once'
1728 cfg
.emailfreq
= 2; // '-M daily'
1731 // Start self-test regex check now if time was not read from state file
1732 if (!cfg
.test_regex
.empty() && !state
.scheduled_test_next_check
)
1733 state
.scheduled_test_next_check
= time(0);
1736 // Common function to format result message for ATA setting
1737 static void format_set_result_msg(std::string
& msg
, const char * name
, bool ok
,
1738 int set_option
= 0, bool has_value
= false)
1745 else if (set_option
< 0)
1748 msg
+= strprintf(":%d", set_option
-1);
1749 else if (set_option
> 0)
1753 // Return true and print message if CFG.dev_idinfo is already in PREV_CFGS
1754 static bool is_duplicate_dev_idinfo(const dev_config
& cfg
, const dev_config_vector
& prev_cfgs
)
1756 if (!cfg
.id_is_unique
)
1759 for (unsigned i
= 0; i
< prev_cfgs
.size(); i
++) {
1760 if (!prev_cfgs
[i
].id_is_unique
)
1762 if (cfg
.dev_idinfo
!= prev_cfgs
[i
].dev_idinfo
.c_str())
1765 PrintOut(LOG_INFO
, "Device: %s, same identity as %s, ignored\n",
1766 cfg
.dev_name
.c_str(), prev_cfgs
[i
].dev_name
.c_str());
1773 // TODO: Add '-F swapid' directive
1774 const bool fix_swapped_id
= false;
1776 // scan to see what ata devices there are, and if they support SMART
1777 static int ATADeviceScan(dev_config
& cfg
, dev_state
& state
, ata_device
* atadev
,
1778 const dev_config_vector
* prev_cfgs
)
1781 struct ata_identify_device drive
;
1782 const char *name
= cfg
.name
.c_str();
1785 // Device must be open
1787 // Get drive identity structure
1788 if ((retid
= ata_read_identity(atadev
, &drive
, fix_swapped_id
))) {
1790 // Unable to read Identity structure
1791 PrintOut(LOG_INFO
,"Device: %s, not ATA, no IDENTIFY DEVICE Structure\n",name
);
1793 PrintOut(LOG_INFO
,"Device: %s, packet devices [this device %s] not SMART capable\n",
1794 name
, packetdevicetype(retid
-1));
1795 CloseDevice(atadev
, name
);
1799 // Get drive identity, size and rotation rate (HDD/SSD)
1800 char model
[40+1], serial
[20+1], firmware
[8+1];
1801 ata_format_id_string(model
, drive
.model
, sizeof(model
)-1);
1802 ata_format_id_string(serial
, drive
.serial_no
, sizeof(serial
)-1);
1803 ata_format_id_string(firmware
, drive
.fw_rev
, sizeof(firmware
)-1);
1805 ata_size_info sizes
;
1806 ata_get_size_info(&drive
, sizes
);
1807 state
.num_sectors
= sizes
.sectors
;
1808 cfg
.dev_rpm
= ata_get_rotation_rate(&drive
);
1810 char wwn
[30]; wwn
[0] = 0;
1811 unsigned oui
= 0; uint64_t unique_id
= 0;
1812 int naa
= ata_get_wwn(&drive
, oui
, unique_id
);
1814 snprintf(wwn
, sizeof(wwn
), "WWN:%x-%06x-%09" PRIx64
", ", naa
, oui
, unique_id
);
1816 // Format device id string for warning emails
1818 cfg
.dev_idinfo
= strprintf("%s, S/N:%s, %sFW:%s, %s", model
, serial
, wwn
, firmware
,
1819 format_capacity(cap
, sizeof(cap
), sizes
.capacity
, "."));
1820 cfg
.id_is_unique
= true; // TODO: Check serial?
1822 PrintOut(LOG_INFO
, "Device: %s, %s\n", name
, cfg
.dev_idinfo
.c_str());
1824 // Check for duplicates
1825 if (prev_cfgs
&& is_duplicate_dev_idinfo(cfg
, *prev_cfgs
)) {
1826 CloseDevice(atadev
, name
);
1830 // Show if device in database, and use preset vendor attribute
1831 // options unless user has requested otherwise.
1832 if (cfg
.ignorepresets
)
1833 PrintOut(LOG_INFO
, "Device: %s, smartd database not searched (Directive: -P ignore).\n", name
);
1835 // Apply vendor specific presets, print warning if present
1836 const drive_settings
* dbentry
= lookup_drive_apply_presets(
1837 &drive
, cfg
.attribute_defs
, cfg
.firmwarebugs
);
1839 PrintOut(LOG_INFO
, "Device: %s, not found in smartd database.\n", name
);
1841 PrintOut(LOG_INFO
, "Device: %s, found in smartd database%s%s\n",
1842 name
, (*dbentry
->modelfamily
? ": " : "."), (*dbentry
->modelfamily
? dbentry
->modelfamily
: ""));
1843 if (*dbentry
->warningmsg
)
1844 PrintOut(LOG_CRIT
, "Device: %s, WARNING: %s\n", name
, dbentry
->warningmsg
);
1848 // Check for ATA Security LOCK
1849 unsigned short word128
= drive
.words088_255
[128-88];
1850 bool locked
= ((word128
& 0x0007) == 0x0007); // LOCKED|ENABLED|SUPPORTED
1852 PrintOut(LOG_INFO
, "Device: %s, ATA Security is **LOCKED**\n", name
);
1854 // Set default '-C 197[+]' if no '-C ID' is specified.
1855 if (!cfg
.curr_pending_set
)
1856 cfg
.curr_pending_id
= get_unc_attr_id(false, cfg
.attribute_defs
, cfg
.curr_pending_incr
);
1857 // Set default '-U 198[+]' if no '-U ID' is specified.
1858 if (!cfg
.offl_pending_set
)
1859 cfg
.offl_pending_id
= get_unc_attr_id(true, cfg
.attribute_defs
, cfg
.offl_pending_incr
);
1861 // If requested, show which presets would be used for this drive
1862 if (cfg
.showpresets
) {
1863 int savedebugmode
=debugmode
;
1864 PrintOut(LOG_INFO
, "Device %s: presets are:\n", name
);
1867 show_presets(&drive
);
1868 debugmode
=savedebugmode
;
1871 // see if drive supports SMART
1872 supported
=ataSmartSupport(&drive
);
1875 // drive does NOT support SMART
1876 PrintOut(LOG_INFO
,"Device: %s, lacks SMART capability\n",name
);
1878 // can't tell if drive supports SMART
1879 PrintOut(LOG_INFO
,"Device: %s, ATA IDENTIFY DEVICE words 82-83 don't specify if SMART capable.\n",name
);
1881 // should we proceed anyway?
1882 if (cfg
.permissive
) {
1883 PrintOut(LOG_INFO
,"Device: %s, proceeding since '-T permissive' Directive given.\n",name
);
1886 PrintOut(LOG_INFO
,"Device: %s, to proceed anyway, use '-T permissive' Directive.\n",name
);
1887 CloseDevice(atadev
, name
);
1892 if (ataEnableSmart(atadev
)) {
1893 // Enable SMART command has failed
1894 PrintOut(LOG_INFO
,"Device: %s, could not enable SMART capability\n",name
);
1896 if (ataIsSmartEnabled(&drive
) <= 0) {
1897 CloseDevice(atadev
, name
);
1900 PrintOut(LOG_INFO
, "Device: %s, proceeding since SMART is already enabled\n", name
);
1903 // disable device attribute autosave...
1904 if (cfg
.autosave
==1) {
1905 if (ataDisableAutoSave(atadev
))
1906 PrintOut(LOG_INFO
,"Device: %s, could not disable SMART Attribute Autosave.\n",name
);
1908 PrintOut(LOG_INFO
,"Device: %s, disabled SMART Attribute Autosave.\n",name
);
1911 // or enable device attribute autosave
1912 if (cfg
.autosave
==2) {
1913 if (ataEnableAutoSave(atadev
))
1914 PrintOut(LOG_INFO
,"Device: %s, could not enable SMART Attribute Autosave.\n",name
);
1916 PrintOut(LOG_INFO
,"Device: %s, enabled SMART Attribute Autosave.\n",name
);
1919 // capability check: SMART status
1920 if (cfg
.smartcheck
&& ataSmartStatus2(atadev
) == -1) {
1921 PrintOut(LOG_INFO
,"Device: %s, not capable of SMART Health Status check\n",name
);
1922 cfg
.smartcheck
= false;
1925 // capability check: Read smart values and thresholds. Note that
1926 // smart values are ALSO needed even if we ONLY want to know if the
1927 // device is self-test log or error-log capable! After ATA-5, this
1928 // information was ALSO reproduced in the IDENTIFY DEVICE response,
1929 // but sadly not for ATA-5. Sigh.
1931 // do we need to get SMART data?
1932 bool smart_val_ok
= false;
1933 if ( cfg
.autoofflinetest
|| cfg
.selftest
1934 || cfg
.errorlog
|| cfg
.xerrorlog
1935 || cfg
.offlinests
|| cfg
.selfteststs
1936 || cfg
.usagefailed
|| cfg
.prefail
|| cfg
.usage
1937 || cfg
.tempdiff
|| cfg
.tempinfo
|| cfg
.tempcrit
1938 || cfg
.curr_pending_id
|| cfg
.offl_pending_id
) {
1940 if (ataReadSmartValues(atadev
, &state
.smartval
)) {
1941 PrintOut(LOG_INFO
, "Device: %s, Read SMART Values failed\n", name
);
1942 cfg
.usagefailed
= cfg
.prefail
= cfg
.usage
= false;
1943 cfg
.tempdiff
= cfg
.tempinfo
= cfg
.tempcrit
= 0;
1944 cfg
.curr_pending_id
= cfg
.offl_pending_id
= 0;
1947 smart_val_ok
= true;
1948 if (ataReadSmartThresholds(atadev
, &state
.smartthres
)) {
1949 PrintOut(LOG_INFO
, "Device: %s, Read SMART Thresholds failed%s\n",
1950 name
, (cfg
.usagefailed
? ", ignoring -f Directive" : ""));
1951 cfg
.usagefailed
= false;
1952 // Let ata_get_attr_state() return ATTRSTATE_NO_THRESHOLD:
1953 memset(&state
.smartthres
, 0, sizeof(state
.smartthres
));
1957 // see if the necessary Attribute is there to monitor offline or
1958 // current pending sectors or temperature
1959 if ( cfg
.curr_pending_id
1960 && !check_pending_id(cfg
, state
, cfg
.curr_pending_id
,
1961 "Current_Pending_Sector"))
1962 cfg
.curr_pending_id
= 0;
1964 if ( cfg
.offl_pending_id
1965 && !check_pending_id(cfg
, state
, cfg
.offl_pending_id
,
1966 "Offline_Uncorrectable"))
1967 cfg
.offl_pending_id
= 0;
1969 if ( (cfg
.tempdiff
|| cfg
.tempinfo
|| cfg
.tempcrit
)
1970 && !ata_return_temperature_value(&state
.smartval
, cfg
.attribute_defs
)) {
1971 PrintOut(LOG_INFO
, "Device: %s, can't monitor Temperature, ignoring -W %d,%d,%d\n",
1972 name
, cfg
.tempdiff
, cfg
.tempinfo
, cfg
.tempcrit
);
1973 cfg
.tempdiff
= cfg
.tempinfo
= cfg
.tempcrit
= 0;
1976 // Report ignored '-r' or '-R' directives
1977 for (int id
= 1; id
<= 255; id
++) {
1978 if (cfg
.monitor_attr_flags
.is_set(id
, MONITOR_RAW_PRINT
)) {
1979 char opt
= (!cfg
.monitor_attr_flags
.is_set(id
, MONITOR_RAW
) ? 'r' : 'R');
1980 const char * excl
= (cfg
.monitor_attr_flags
.is_set(id
,
1981 (opt
== 'r' ? MONITOR_AS_CRIT
: MONITOR_RAW_AS_CRIT
)) ? "!" : "");
1983 int idx
= ata_find_attr_index(id
, state
.smartval
);
1985 PrintOut(LOG_INFO
,"Device: %s, no Attribute %d, ignoring -%c %d%s\n", name
, id
, opt
, id
, excl
);
1987 bool prefail
= !!ATTRIBUTE_FLAGS_PREFAILURE(state
.smartval
.vendor_attributes
[idx
].flags
);
1988 if (!((prefail
&& cfg
.prefail
) || (!prefail
&& cfg
.usage
)))
1989 PrintOut(LOG_INFO
,"Device: %s, not monitoring %s Attributes, ignoring -%c %d%s\n", name
,
1990 (prefail
? "Prefailure" : "Usage"), opt
, id
, excl
);
1996 // enable/disable automatic on-line testing
1997 if (cfg
.autoofflinetest
) {
1998 // is this an enable or disable request?
1999 const char *what
=(cfg
.autoofflinetest
==1)?"disable":"enable";
2001 PrintOut(LOG_INFO
,"Device: %s, could not %s SMART Automatic Offline Testing.\n",name
, what
);
2003 // if command appears unsupported, issue a warning...
2004 if (!isSupportAutomaticTimer(&state
.smartval
))
2005 PrintOut(LOG_INFO
,"Device: %s, SMART Automatic Offline Testing unsupported...\n",name
);
2006 // ... but then try anyway
2007 if ((cfg
.autoofflinetest
==1)?ataDisableAutoOffline(atadev
):ataEnableAutoOffline(atadev
))
2008 PrintOut(LOG_INFO
,"Device: %s, %s SMART Automatic Offline Testing failed.\n", name
, what
);
2010 PrintOut(LOG_INFO
,"Device: %s, %sd SMART Automatic Offline Testing.\n", name
, what
);
2014 // Read log directories if required for capability check
2015 ata_smart_log_directory smart_logdir
, gp_logdir
;
2016 bool smart_logdir_ok
= false, gp_logdir_ok
= false;
2018 if ( isGeneralPurposeLoggingCapable(&drive
)
2019 && (cfg
.errorlog
|| cfg
.selftest
)
2020 && !cfg
.firmwarebugs
.is_set(BUG_NOLOGDIR
)) {
2021 if (!ataReadLogDirectory(atadev
, &smart_logdir
, false))
2022 smart_logdir_ok
= true;
2025 if (cfg
.xerrorlog
&& !cfg
.firmwarebugs
.is_set(BUG_NOLOGDIR
)) {
2026 if (!ataReadLogDirectory(atadev
, &gp_logdir
, true))
2027 gp_logdir_ok
= true;
2030 // capability check: self-test-log
2031 state
.selflogcount
= 0; state
.selfloghour
= 0;
2034 if (!( cfg
.permissive
2035 || ( smart_logdir_ok
&& smart_logdir
.entry
[0x06-1].numsectors
)
2036 || (!smart_logdir_ok
&& smart_val_ok
&& isSmartTestLogCapable(&state
.smartval
, &drive
)))) {
2037 PrintOut(LOG_INFO
, "Device: %s, no SMART Self-test Log, ignoring -l selftest (override with -T permissive)\n", name
);
2038 cfg
.selftest
= false;
2040 else if ((retval
= SelfTestErrorCount(atadev
, name
, cfg
.firmwarebugs
)) < 0) {
2041 PrintOut(LOG_INFO
, "Device: %s, no SMART Self-test Log, ignoring -l selftest\n", name
);
2042 cfg
.selftest
= false;
2045 state
.selflogcount
=SELFTEST_ERRORCOUNT(retval
);
2046 state
.selfloghour
=SELFTEST_ERRORHOURS(retval
);
2050 // capability check: ATA error log
2051 state
.ataerrorcount
= 0;
2054 if (!( cfg
.permissive
2055 || ( smart_logdir_ok
&& smart_logdir
.entry
[0x01-1].numsectors
)
2056 || (!smart_logdir_ok
&& smart_val_ok
&& isSmartErrorLogCapable(&state
.smartval
, &drive
)))) {
2057 PrintOut(LOG_INFO
, "Device: %s, no SMART Error Log, ignoring -l error (override with -T permissive)\n", name
);
2058 cfg
.errorlog
= false;
2060 else if ((errcnt1
= read_ata_error_count(atadev
, name
, cfg
.firmwarebugs
, false)) < 0) {
2061 PrintOut(LOG_INFO
, "Device: %s, no SMART Error Log, ignoring -l error\n", name
);
2062 cfg
.errorlog
= false;
2065 state
.ataerrorcount
= errcnt1
;
2068 if (cfg
.xerrorlog
) {
2070 if (!( cfg
.permissive
|| cfg
.firmwarebugs
.is_set(BUG_NOLOGDIR
)
2071 || (gp_logdir_ok
&& gp_logdir
.entry
[0x03-1].numsectors
) )) {
2072 PrintOut(LOG_INFO
, "Device: %s, no Extended Comprehensive SMART Error Log, ignoring -l xerror (override with -T permissive)\n",
2074 cfg
.xerrorlog
= false;
2076 else if ((errcnt2
= read_ata_error_count(atadev
, name
, cfg
.firmwarebugs
, true)) < 0) {
2077 PrintOut(LOG_INFO
, "Device: %s, no Extended Comprehensive SMART Error Log, ignoring -l xerror\n", name
);
2078 cfg
.xerrorlog
= false;
2080 else if (cfg
.errorlog
&& state
.ataerrorcount
!= errcnt2
) {
2081 PrintOut(LOG_INFO
, "Device: %s, SMART Error Logs report different error counts: %d != %d\n",
2082 name
, state
.ataerrorcount
, errcnt2
);
2083 // Record max error count
2084 if (errcnt2
> state
.ataerrorcount
)
2085 state
.ataerrorcount
= errcnt2
;
2088 state
.ataerrorcount
= errcnt2
;
2091 // capability check: self-test and offline data collection status
2092 if (cfg
.offlinests
|| cfg
.selfteststs
) {
2093 if (!(cfg
.permissive
|| (smart_val_ok
&& state
.smartval
.offline_data_collection_capability
))) {
2095 PrintOut(LOG_INFO
, "Device: %s, no SMART Offline Data Collection capability, ignoring -l offlinests (override with -T permissive)\n", name
);
2096 if (cfg
.selfteststs
)
2097 PrintOut(LOG_INFO
, "Device: %s, no SMART Self-test capability, ignoring -l selfteststs (override with -T permissive)\n", name
);
2098 cfg
.offlinests
= cfg
.selfteststs
= false;
2102 // capabilities check -- does it support powermode?
2103 if (cfg
.powermode
) {
2104 int powermode
= ataCheckPowerMode(atadev
);
2106 if (-1 == powermode
) {
2107 PrintOut(LOG_CRIT
, "Device: %s, no ATA CHECK POWER STATUS support, ignoring -n Directive\n", name
);
2110 else if (powermode
!=0x00 && powermode
!=0x01
2111 && powermode
!=0x40 && powermode
!=0x41
2112 && powermode
!=0x80 && powermode
!=0x81 && powermode
!=0x82 && powermode
!=0x83
2113 && powermode
!=0xff) {
2114 PrintOut(LOG_CRIT
, "Device: %s, CHECK POWER STATUS returned %d, not ATA compliant, ignoring -n Directive\n",
2120 // Apply ATA settings
2124 format_set_result_msg(msg
, "AAM", (cfg
.set_aam
> 0 ?
2125 ata_set_features(atadev
, ATA_ENABLE_AAM
, cfg
.set_aam
-1) :
2126 ata_set_features(atadev
, ATA_DISABLE_AAM
)), cfg
.set_aam
, true);
2129 format_set_result_msg(msg
, "APM", (cfg
.set_apm
> 0 ?
2130 ata_set_features(atadev
, ATA_ENABLE_APM
, cfg
.set_apm
-1) :
2131 ata_set_features(atadev
, ATA_DISABLE_APM
)), cfg
.set_apm
, true);
2133 if (cfg
.set_lookahead
)
2134 format_set_result_msg(msg
, "Rd-ahead", ata_set_features(atadev
,
2135 (cfg
.set_lookahead
> 0 ? ATA_ENABLE_READ_LOOK_AHEAD
: ATA_DISABLE_READ_LOOK_AHEAD
)),
2139 format_set_result_msg(msg
, "Wr-cache", ata_set_features(atadev
,
2140 (cfg
.set_wcache
> 0? ATA_ENABLE_WRITE_CACHE
: ATA_DISABLE_WRITE_CACHE
)), cfg
.set_wcache
);
2143 format_set_result_msg(msg
, "DSN", ata_set_features(atadev
,
2144 ATA_ENABLE_DISABLE_DSN
, (cfg
.set_dsn
> 0 ? 0x1 : 0x2)));
2146 if (cfg
.set_security_freeze
)
2147 format_set_result_msg(msg
, "Security freeze",
2148 ata_nodata_command(atadev
, ATA_SECURITY_FREEZE_LOCK
));
2150 if (cfg
.set_standby
)
2151 format_set_result_msg(msg
, "Standby",
2152 ata_nodata_command(atadev
, ATA_IDLE
, cfg
.set_standby
-1), cfg
.set_standby
, true);
2154 // Report as one log entry
2156 PrintOut(LOG_INFO
, "Device: %s, ATA settings applied: %s\n", name
, msg
.c_str());
2158 // set SCT Error Recovery Control if requested
2159 if (cfg
.sct_erc_set
) {
2160 if (!isSCTErrorRecoveryControlCapable(&drive
))
2161 PrintOut(LOG_INFO
, "Device: %s, no SCT Error Recovery Control support, ignoring -l scterc\n",
2164 PrintOut(LOG_INFO
, "Device: %s, no SCT support if ATA Security is LOCKED, ignoring -l scterc\n",
2166 else if ( ataSetSCTErrorRecoveryControltime(atadev
, 1, cfg
.sct_erc_readtime
)
2167 || ataSetSCTErrorRecoveryControltime(atadev
, 2, cfg
.sct_erc_writetime
))
2168 PrintOut(LOG_INFO
, "Device: %s, set of SCT Error Recovery Control failed\n", name
);
2170 PrintOut(LOG_INFO
, "Device: %s, SCT Error Recovery Control set to: Read: %u, Write: %u\n",
2171 name
, cfg
.sct_erc_readtime
, cfg
.sct_erc_writetime
);
2174 // If no tests available or selected, return
2175 if (!( cfg
.smartcheck
|| cfg
.selftest
2176 || cfg
.errorlog
|| cfg
.xerrorlog
2177 || cfg
.offlinests
|| cfg
.selfteststs
2178 || cfg
.usagefailed
|| cfg
.prefail
|| cfg
.usage
2179 || cfg
.tempdiff
|| cfg
.tempinfo
|| cfg
.tempcrit
)) {
2180 CloseDevice(atadev
, name
);
2184 // tell user we are registering device
2185 PrintOut(LOG_INFO
,"Device: %s, is SMART capable. Adding to \"monitor\" list.\n",name
);
2187 // close file descriptor
2188 CloseDevice(atadev
, name
);
2190 if (!state_path_prefix
.empty() || !attrlog_path_prefix
.empty()) {
2191 // Build file name for state file
2192 std::replace_if(model
, model
+strlen(model
), not_allowed_in_filename
, '_');
2193 std::replace_if(serial
, serial
+strlen(serial
), not_allowed_in_filename
, '_');
2194 if (!state_path_prefix
.empty()) {
2195 cfg
.state_file
= strprintf("%s%s-%s.ata.state", state_path_prefix
.c_str(), model
, serial
);
2196 // Read previous state
2197 if (read_dev_state(cfg
.state_file
.c_str(), state
)) {
2198 PrintOut(LOG_INFO
, "Device: %s, state read from %s\n", name
, cfg
.state_file
.c_str());
2199 // Copy ATA attribute values to temp state
2200 state
.update_temp_state();
2203 if (!attrlog_path_prefix
.empty())
2204 cfg
.attrlog_file
= strprintf("%s%s-%s.ata.csv", attrlog_path_prefix
.c_str(), model
, serial
);
2207 finish_device_scan(cfg
, state
);
2212 // on success, return 0. On failure, return >0. Never return <0,
2214 static int SCSIDeviceScan(dev_config
& cfg
, dev_state
& state
, scsi_device
* scsidev
,
2215 const dev_config_vector
* prev_cfgs
)
2217 int err
, req_len
, avail_len
, version
, len
;
2218 const char *device
= cfg
.name
.c_str();
2219 struct scsi_iec_mode_page iec
;
2223 char lu_id
[64], serial
[256], vendor
[40], model
[40];
2225 // Device must be open
2226 memset(inqBuf
, 0, 96);
2228 if ((err
= scsiStdInquiry(scsidev
, inqBuf
, req_len
))) {
2229 /* Marvell controllers fail on a 36 bytes StdInquiry, but 64 suffices */
2231 if ((err
= scsiStdInquiry(scsidev
, inqBuf
, req_len
))) {
2232 PrintOut(LOG_INFO
, "Device: %s, Both 36 and 64 byte INQUIRY failed; "
2233 "skip device\n", device
);
2237 version
= (inqBuf
[2] & 0x7f); /* Accept old ISO/IEC 9316:1995 variants */
2239 avail_len
= inqBuf
[4] + 5;
2240 len
= (avail_len
< req_len
) ? avail_len
: req_len
;
2242 PrintOut(LOG_INFO
, "Device: %s, INQUIRY response less than 36 bytes; "
2243 "skip device\n", device
);
2247 int pdt
= inqBuf
[0] & 0x1f;
2249 if (! ((0 == pdt
) || (4 == pdt
) || (5 == pdt
) || (7 == pdt
) ||
2251 PrintOut(LOG_INFO
, "Device: %s, not a disk like device [PDT=0x%x], "
2252 "skip\n", device
, pdt
);
2256 if (supported_vpd_pages_p
) {
2257 delete supported_vpd_pages_p
;
2258 supported_vpd_pages_p
= NULL
;
2260 supported_vpd_pages_p
= new supported_vpd_pages(scsidev
);
2263 if ((version
>= 0x3) && (version
< 0x8)) {
2265 if (0 == scsiInquiryVpd(scsidev
, SCSI_VPD_DEVICE_IDENTIFICATION
,
2266 vpdBuf
, sizeof(vpdBuf
))) {
2268 scsi_decode_lu_dev_id(vpdBuf
+ 4, len
, lu_id
, sizeof(lu_id
), NULL
);
2272 if (0 == scsiInquiryVpd(scsidev
, SCSI_VPD_UNIT_SERIAL_NUMBER
,
2273 vpdBuf
, sizeof(vpdBuf
))) {
2275 vpdBuf
[4 + len
] = '\0';
2276 scsi_format_id_string(serial
, (const unsigned char *)&vpdBuf
[4], len
);
2279 unsigned int lb_size
;
2281 uint64_t capacity
= scsiGetSize(scsidev
, &lb_size
, NULL
);
2284 format_capacity(si_str
, sizeof(si_str
), capacity
, ".");
2288 // Format device id string for warning emails
2289 cfg
.dev_idinfo
= strprintf("[%.8s %.16s %.4s]%s%s%s%s%s%s",
2290 (char *)&inqBuf
[8], (char *)&inqBuf
[16], (char *)&inqBuf
[32],
2291 (lu_id
[0] ? ", lu id: " : ""), (lu_id
[0] ? lu_id
: ""),
2292 (serial
[0] ? ", S/N: " : ""), (serial
[0] ? serial
: ""),
2293 (si_str
[0] ? ", " : ""), (si_str
[0] ? si_str
: ""));
2294 cfg
.id_is_unique
= (lu_id
[0] || serial
[0]);
2296 // format "model" string
2297 scsi_format_id_string(vendor
, (const unsigned char *)&inqBuf
[8], 8);
2298 scsi_format_id_string(model
, (const unsigned char *)&inqBuf
[16], 16);
2299 PrintOut(LOG_INFO
, "Device: %s, %s\n", device
, cfg
.dev_idinfo
.c_str());
2301 // Check for duplicates
2302 if (prev_cfgs
&& is_duplicate_dev_idinfo(cfg
, *prev_cfgs
)) {
2303 CloseDevice(scsidev
, device
);
2307 // check that device is ready for commands. IE stores its stuff on
2309 if ((err
= scsiTestUnitReady(scsidev
))) {
2310 if (SIMPLE_ERR_NOT_READY
== err
)
2311 PrintOut(LOG_INFO
, "Device: %s, NOT READY (e.g. spun down); skip device\n", device
);
2312 else if (SIMPLE_ERR_NO_MEDIUM
== err
)
2313 PrintOut(LOG_INFO
, "Device: %s, NO MEDIUM present; skip device\n", device
);
2314 else if (SIMPLE_ERR_BECOMING_READY
== err
)
2315 PrintOut(LOG_INFO
, "Device: %s, BECOMING (but not yet) READY; skip device\n", device
);
2317 PrintOut(LOG_CRIT
, "Device: %s, failed Test Unit Ready [err=%d]\n", device
, err
);
2318 CloseDevice(scsidev
, device
);
2322 // Badly-conforming USB storage devices may fail this check.
2323 // The response to the following IE mode page fetch (current and
2324 // changeable values) is carefully examined. It has been found
2325 // that various USB devices that malform the response will lock up
2326 // if asked for a log page (e.g. temperature) so it is best to
2328 if (!(err
= scsiFetchIECmpage(scsidev
, &iec
, state
.modese_len
)))
2329 state
.modese_len
= iec
.modese_len
;
2330 else if (SIMPLE_ERR_BAD_FIELD
== err
)
2331 ; /* continue since it is reasonable not to support IE mpage */
2332 else { /* any other error (including malformed response) unreasonable */
2334 "Device: %s, Bad IEC (SMART) mode page, err=%d, skip device\n",
2336 CloseDevice(scsidev
, device
);
2340 // N.B. The following is passive (i.e. it doesn't attempt to turn on
2341 // smart if it is off). This may change to be the same as the ATA side.
2342 if (!scsi_IsExceptionControlEnabled(&iec
)) {
2343 PrintOut(LOG_INFO
, "Device: %s, IE (SMART) not enabled, skip device\n"
2344 "Try 'smartctl -s on %s' to turn on SMART features\n",
2346 CloseDevice(scsidev
, device
);
2350 // Flag that certain log pages are supported (information may be
2351 // available from other sources).
2352 if (0 == scsiLogSense(scsidev
, SUPPORTED_LPAGES
, 0, tBuf
, sizeof(tBuf
), 0) ||
2353 0 == scsiLogSense(scsidev
, SUPPORTED_LPAGES
, 0, tBuf
, sizeof(tBuf
), 68))
2354 /* workaround for the bug #678 on ST8000NM0075/E001. Up to 64 pages + 4b header */
2356 for (int k
= 4; k
< tBuf
[3] + LOGPAGEHDRSIZE
; ++k
) {
2358 case TEMPERATURE_LPAGE
:
2359 state
.TempPageSupported
= 1;
2362 state
.SmartPageSupported
= 1;
2364 case READ_ERROR_COUNTER_LPAGE
:
2365 state
.ReadECounterPageSupported
= 1;
2367 case WRITE_ERROR_COUNTER_LPAGE
:
2368 state
.WriteECounterPageSupported
= 1;
2370 case VERIFY_ERROR_COUNTER_LPAGE
:
2371 state
.VerifyECounterPageSupported
= 1;
2373 case NON_MEDIUM_ERROR_LPAGE
:
2374 state
.NonMediumErrorPageSupported
= 1;
2382 // Check if scsiCheckIE() is going to work
2386 UINT8 currenttemp
= 0;
2389 if (scsiCheckIE(scsidev
, state
.SmartPageSupported
, state
.TempPageSupported
,
2390 &asc
, &ascq
, ¤ttemp
, &triptemp
)) {
2391 PrintOut(LOG_INFO
, "Device: %s, unexpectedly failed to read SMART values\n", device
);
2392 state
.SuppressReport
= 1;
2394 if ( (state
.SuppressReport
|| !currenttemp
)
2395 && (cfg
.tempdiff
|| cfg
.tempinfo
|| cfg
.tempcrit
)) {
2396 PrintOut(LOG_INFO
, "Device: %s, can't monitor Temperature, ignoring -W %d,%d,%d\n",
2397 device
, cfg
.tempdiff
, cfg
.tempinfo
, cfg
.tempcrit
);
2398 cfg
.tempdiff
= cfg
.tempinfo
= cfg
.tempcrit
= 0;
2402 // capability check: self-test-log
2404 int retval
= scsiCountFailedSelfTests(scsidev
, 0);
2406 // no self-test log, turn off monitoring
2407 PrintOut(LOG_INFO
, "Device: %s, does not support SMART Self-Test Log.\n", device
);
2408 cfg
.selftest
= false;
2409 state
.selflogcount
= 0;
2410 state
.selfloghour
= 0;
2413 // register starting values to watch for changes
2414 state
.selflogcount
=SELFTEST_ERRORCOUNT(retval
);
2415 state
.selfloghour
=SELFTEST_ERRORHOURS(retval
);
2419 // disable autosave (set GLTSD bit)
2420 if (cfg
.autosave
==1){
2421 if (scsiSetControlGLTSD(scsidev
, 1, state
.modese_len
))
2422 PrintOut(LOG_INFO
,"Device: %s, could not disable autosave (set GLTSD bit).\n",device
);
2424 PrintOut(LOG_INFO
,"Device: %s, disabled autosave (set GLTSD bit).\n",device
);
2427 // or enable autosave (clear GLTSD bit)
2428 if (cfg
.autosave
==2){
2429 if (scsiSetControlGLTSD(scsidev
, 0, state
.modese_len
))
2430 PrintOut(LOG_INFO
,"Device: %s, could not enable autosave (clear GLTSD bit).\n",device
);
2432 PrintOut(LOG_INFO
,"Device: %s, enabled autosave (cleared GLTSD bit).\n",device
);
2435 // tell user we are registering device
2436 PrintOut(LOG_INFO
, "Device: %s, is SMART capable. Adding to \"monitor\" list.\n", device
);
2438 // Make sure that init_standby_check() ignores SCSI devices
2439 cfg
.offlinests_ns
= cfg
.selfteststs_ns
= false;
2441 // close file descriptor
2442 CloseDevice(scsidev
, device
);
2444 if (!state_path_prefix
.empty() || !attrlog_path_prefix
.empty()) {
2445 // Build file name for state file
2446 std::replace_if(model
, model
+strlen(model
), not_allowed_in_filename
, '_');
2447 std::replace_if(serial
, serial
+strlen(serial
), not_allowed_in_filename
, '_');
2448 if (!state_path_prefix
.empty()) {
2449 cfg
.state_file
= strprintf("%s%s-%s-%s.scsi.state", state_path_prefix
.c_str(), vendor
, model
, serial
);
2450 // Read previous state
2451 if (read_dev_state(cfg
.state_file
.c_str(), state
)) {
2452 PrintOut(LOG_INFO
, "Device: %s, state read from %s\n", device
, cfg
.state_file
.c_str());
2453 // Copy ATA attribute values to temp state
2454 state
.update_temp_state();
2457 if (!attrlog_path_prefix
.empty())
2458 cfg
.attrlog_file
= strprintf("%s%s-%s-%s.scsi.csv", attrlog_path_prefix
.c_str(), vendor
, model
, serial
);
2461 finish_device_scan(cfg
, state
);
2466 // Convert 128 bit LE integer to uint64_t or its max value on overflow.
2467 static uint64_t le128_to_uint64(const unsigned char (& val
)[16])
2469 for (int i
= 8; i
< 16; i
++) {
2471 return ~(uint64_t)0;
2473 uint64_t lo
= val
[7];
2474 for (int i
= 7-1; i
>= 0; i
--) {
2475 lo
<<= 8; lo
+= val
[i
];
2480 // Get max temperature in Kelvin reported in NVMe SMART/Health log.
2481 static int nvme_get_max_temp_kelvin(const nvme_smart_log
& smart_log
)
2483 int k
= (smart_log
.temperature
[1] << 8) | smart_log
.temperature
[0];
2484 for (int i
= 0; i
< 8; i
++) {
2485 if (smart_log
.temp_sensor
[i
] > k
)
2486 k
= smart_log
.temp_sensor
[i
];
2491 static int NVMeDeviceScan(dev_config
& cfg
, dev_state
& state
, nvme_device
* nvmedev
,
2492 const dev_config_vector
* prev_cfgs
)
2494 const char *name
= cfg
.name
.c_str();
2496 // Device must be open
2498 // Get ID Controller
2499 nvme_id_ctrl id_ctrl
;
2500 if (!nvme_read_id_ctrl(nvmedev
, id_ctrl
)) {
2501 PrintOut(LOG_INFO
, "Device: %s, NVMe Identify Controller failed\n", name
);
2502 CloseDevice(nvmedev
, name
);
2506 // Get drive identity
2507 char model
[40+1], serial
[20+1], firmware
[8+1];
2508 format_char_array(model
, id_ctrl
.mn
);
2509 format_char_array(serial
, id_ctrl
.sn
);
2510 format_char_array(firmware
, id_ctrl
.fr
);
2512 // Format device id string for warning emails
2513 char nsstr
[32] = "", capstr
[32] = "";
2514 unsigned nsid
= nvmedev
->get_nsid();
2515 if (nsid
!= 0xffffffff)
2516 snprintf(nsstr
, sizeof(nsstr
), ", NSID:%u", nsid
);
2517 uint64_t capacity
= le128_to_uint64(id_ctrl
.tnvmcap
);
2519 format_capacity(capstr
, sizeof(capstr
), capacity
, ".");
2520 cfg
.dev_idinfo
= strprintf("%s, S/N:%s, FW:%s%s%s%s", model
, serial
, firmware
,
2521 nsstr
, (capstr
[0] ? ", " : ""), capstr
);
2522 cfg
.id_is_unique
= true; // TODO: Check serial?
2524 PrintOut(LOG_INFO
, "Device: %s, %s\n", name
, cfg
.dev_idinfo
.c_str());
2526 // Check for duplicates
2527 if (prev_cfgs
&& is_duplicate_dev_idinfo(cfg
, *prev_cfgs
)) {
2528 CloseDevice(nvmedev
, name
);
2532 // Read SMART/Health log
2533 nvme_smart_log smart_log
;
2534 if (!nvme_read_smart_log(nvmedev
, smart_log
)) {
2535 PrintOut(LOG_INFO
, "Device: %s, failed to read NVMe SMART/Health Information\n", name
);
2536 CloseDevice(nvmedev
, name
);
2540 // Check temperature sensor support
2541 if (cfg
.tempdiff
|| cfg
.tempinfo
|| cfg
.tempcrit
) {
2542 if (!nvme_get_max_temp_kelvin(smart_log
)) {
2543 PrintOut(LOG_INFO
, "Device: %s, no Temperature sensors, ignoring -W %d,%d,%d\n",
2544 name
, cfg
.tempdiff
, cfg
.tempinfo
, cfg
.tempcrit
);
2545 cfg
.tempdiff
= cfg
.tempinfo
= cfg
.tempcrit
= 0;
2549 // Init total error count
2550 if (cfg
.errorlog
|| cfg
.xerrorlog
) {
2551 state
.nvme_err_log_entries
= le128_to_uint64(smart_log
.num_err_log_entries
);
2554 // If no supported tests selected, return
2555 if (!( cfg
.smartcheck
|| cfg
.errorlog
|| cfg
.xerrorlog
2556 || cfg
.tempdiff
|| cfg
.tempinfo
|| cfg
.tempcrit
)) {
2557 CloseDevice(nvmedev
, name
);
2561 // Tell user we are registering device
2562 PrintOut(LOG_INFO
,"Device: %s, is SMART capable. Adding to \"monitor\" list.\n", name
);
2564 // Make sure that init_standby_check() ignores NVMe devices
2565 cfg
.offlinests_ns
= cfg
.selfteststs_ns
= false;
2567 CloseDevice(nvmedev
, name
);
2569 if (!state_path_prefix
.empty()) {
2570 // Build file name for state file
2571 std::replace_if(model
, model
+strlen(model
), not_allowed_in_filename
, '_');
2572 std::replace_if(serial
, serial
+strlen(serial
), not_allowed_in_filename
, '_');
2574 if (nsid
!= 0xffffffff)
2575 snprintf(nsstr
, sizeof(nsstr
), "-n%u", nsid
);
2576 cfg
.state_file
= strprintf("%s%s-%s%s.nvme.state", state_path_prefix
.c_str(), model
, serial
, nsstr
);
2577 // Read previous state
2578 if (read_dev_state(cfg
.state_file
.c_str(), state
))
2579 PrintOut(LOG_INFO
, "Device: %s, state read from %s\n", name
, cfg
.state_file
.c_str());
2582 finish_device_scan(cfg
, state
);
2587 // Open device for next check, return false on error
2588 static bool open_device(const dev_config
& cfg
, dev_state
& state
, smart_device
* device
,
2591 const char * name
= cfg
.name
.c_str();
2593 // If user has asked, test the email warning system
2595 MailWarning(cfg
, state
, 0, "TEST EMAIL from smartd for device: %s", name
);
2597 // User may have requested (with the -n Directive) to leave the disk
2598 // alone if it is in idle or standby mode. In this case check the
2599 // power mode first before opening the device for full access,
2600 // and exit without check if disk is reported in standby.
2601 if (device
->is_ata() && cfg
.powermode
&& !state
.powermodefail
&& !state
.removed
) {
2602 // Note that 'is_powered_down()' handles opening the device itself, and
2603 // can be used before calling 'open()' (that's the whole point of 'is_powered_down()'!).
2604 if (device
->is_powered_down())
2606 // skip at most powerskipmax checks
2607 if (!cfg
.powerskipmax
|| state
.powerskipcnt
<cfg
.powerskipmax
) {
2608 // report first only except if state has changed, avoid waking up system disk
2609 if ((!state
.powerskipcnt
|| state
.lastpowermodeskipped
!= -1) && !cfg
.powerquiet
) {
2610 PrintOut(LOG_INFO
, "Device: %s, is in %s mode, suspending checks\n", name
, "STANDBY (OS)");
2611 state
.lastpowermodeskipped
= -1;
2613 state
.powerskipcnt
++;
2619 // if we can't open device, fail gracefully rather than hard --
2620 // perhaps the next time around we'll be able to open it
2621 if (!device
->open()) {
2622 // For removable devices, print error message only once and suppress email
2623 if (!cfg
.removable
) {
2624 PrintOut(LOG_INFO
, "Device: %s, open() of %s device failed: %s\n", name
, type
, device
->get_errmsg());
2625 MailWarning(cfg
, state
, 9, "Device: %s, unable to open %s device", name
, type
);
2627 else if (!state
.removed
) {
2628 PrintOut(LOG_INFO
, "Device: %s, removed %s device: %s\n", name
, type
, device
->get_errmsg());
2629 state
.removed
= true;
2632 PrintOut(LOG_INFO
, "Device: %s, %s device still removed: %s\n", name
, type
, device
->get_errmsg());
2637 PrintOut(LOG_INFO
,"Device: %s, opened %s device\n", name
, type
);
2640 reset_warning_mail(cfg
, state
, 9, "open of %s device worked again", type
);
2641 else if (state
.removed
) {
2642 PrintOut(LOG_INFO
, "Device: %s, reconnected %s device\n", name
, type
);
2643 state
.removed
= false;
2649 // If the self-test log has got more self-test errors (or more recent
2650 // self-test errors) recorded, then notify user.
2651 static void CheckSelfTestLogs(const dev_config
& cfg
, dev_state
& state
, int newi
)
2653 const char * name
= cfg
.name
.c_str();
2657 MailWarning(cfg
, state
, 8, "Device: %s, Read SMART Self-Test Log Failed", name
);
2659 reset_warning_mail(cfg
, state
, 8, "Read SMART Self-Test Log worked again");
2661 // old and new error counts
2662 int oldc
=state
.selflogcount
;
2663 int newc
=SELFTEST_ERRORCOUNT(newi
);
2665 // old and new error timestamps in hours
2666 int oldh
=state
.selfloghour
;
2667 int newh
=SELFTEST_ERRORHOURS(newi
);
2670 // increase in error count
2671 PrintOut(LOG_CRIT
, "Device: %s, Self-Test Log error count increased from %d to %d\n",
2673 MailWarning(cfg
, state
, 3, "Device: %s, Self-Test Log error count increased from %d to %d",
2675 state
.must_write
= true;
2677 else if (newc
> 0 && oldh
!= newh
) {
2678 // more recent error
2679 // a 'more recent' error might actually be a smaller hour number,
2680 // if the hour number has wrapped.
2681 // There's still a bug here. You might just happen to run a new test
2682 // exactly 32768 hours after the previous failure, and have run exactly
2683 // 20 tests between the two, in which case smartd will miss the
2685 PrintOut(LOG_CRIT
, "Device: %s, new Self-Test Log error at hour timestamp %d\n",
2687 MailWarning(cfg
, state
, 3, "Device: %s, new Self-Test Log error at hour timestamp %d",
2689 state
.must_write
= true;
2692 // Print info if error entries have disappeared
2693 // or newer successful successful extended self-test exits
2695 PrintOut(LOG_INFO
, "Device: %s, Self-Test Log error count decreased from %d to %d\n",
2698 reset_warning_mail(cfg
, state
, 3, "Self-Test Log does no longer report errors");
2701 // Needed since self-test error count may DECREASE. Hour might
2702 // also have changed.
2703 state
.selflogcount
= newc
;
2704 state
.selfloghour
= newh
;
2709 // Test types, ordered by priority.
2710 static const char test_type_chars
[] = "LncrSCO";
2711 static const unsigned num_test_types
= sizeof(test_type_chars
)-1;
2713 // returns test type if time to do test of type testtype,
2714 // 0 if not time to do test.
2715 static char next_scheduled_test(const dev_config
& cfg
, dev_state
& state
, bool scsi
, time_t usetime
= 0)
2717 // check that self-testing has been requested
2718 if (cfg
.test_regex
.empty())
2721 // Exit if drive not capable of any test
2722 if ( state
.not_cap_long
&& state
.not_cap_short
&&
2723 (scsi
|| (state
.not_cap_conveyance
&& state
.not_cap_offline
)))
2726 // since we are about to call localtime(), be sure glibc is informed
2727 // of any timezone changes we make.
2729 FixGlibcTimeZoneBug();
2731 // Is it time for next check?
2732 time_t now
= (!usetime
? time(0) : usetime
);
2733 if (now
< state
.scheduled_test_next_check
)
2736 // Limit time check interval to 90 days
2737 if (state
.scheduled_test_next_check
+ (3600L*24*90) < now
)
2738 state
.scheduled_test_next_check
= now
- (3600L*24*90);
2740 // Check interval [state.scheduled_test_next_check, now] for scheduled tests
2742 time_t testtime
= 0; int testhour
= 0;
2743 int maxtest
= num_test_types
-1;
2745 for (time_t t
= state
.scheduled_test_next_check
; ; ) {
2746 struct tm
* tms
= localtime(&t
);
2747 // tm_wday is 0 (Sunday) to 6 (Saturday). We use 1 (Monday) to 7 (Sunday).
2748 int weekday
= (tms
->tm_wday
? tms
->tm_wday
: 7);
2749 for (int i
= 0; i
<= maxtest
; i
++) {
2750 // Skip if drive not capable of this test
2751 switch (test_type_chars
[i
]) {
2752 case 'L': if (state
.not_cap_long
) continue; break;
2753 case 'S': if (state
.not_cap_short
) continue; break;
2754 case 'C': if (scsi
|| state
.not_cap_conveyance
) continue; break;
2755 case 'O': if (scsi
|| state
.not_cap_offline
) continue; break;
2757 case 'r': if (scsi
|| state
.not_cap_selective
) continue; break;
2760 // Try match of "T/MM/DD/d/HH"
2762 snprintf(pattern
, sizeof(pattern
), "%c/%02d/%02d/%1d/%02d",
2763 test_type_chars
[i
], tms
->tm_mon
+1, tms
->tm_mday
, weekday
, tms
->tm_hour
);
2764 if (cfg
.test_regex
.full_match(pattern
)) {
2766 testtype
= pattern
[0];
2767 testtime
= t
; testhour
= tms
->tm_hour
;
2768 // Limit further matches to higher priority self-tests
2773 // Exit if no tests left or current time reached
2779 if ((t
+= 3600) > now
)
2783 // Do next check not before next hour.
2784 struct tm
* tmnow
= localtime(&now
);
2785 state
.scheduled_test_next_check
= now
+ (3600 - tmnow
->tm_min
*60 - tmnow
->tm_sec
);
2788 state
.must_write
= true;
2789 // Tell user if an old test was found.
2790 if (!usetime
&& !(testhour
== tmnow
->tm_hour
&& testtime
+ 3600 > now
)) {
2791 char datebuf
[DATEANDEPOCHLEN
]; dateandtimezoneepoch(datebuf
, testtime
);
2792 PrintOut(LOG_INFO
, "Device: %s, old test of type %c not run at %s, starting now.\n",
2793 cfg
.name
.c_str(), testtype
, datebuf
);
2800 // Print a list of future tests.
2801 static void PrintTestSchedule(const dev_config_vector
& configs
, dev_state_vector
& states
, const smart_device_list
& devices
)
2803 unsigned numdev
= configs
.size();
2806 std::vector
<int> testcnts(numdev
* num_test_types
, 0);
2808 PrintOut(LOG_INFO
, "\nNext scheduled self tests (at most 5 of each type per device):\n");
2810 // FixGlibcTimeZoneBug(); // done in PrintOut()
2811 time_t now
= time(0);
2812 char datenow
[DATEANDEPOCHLEN
], date
[DATEANDEPOCHLEN
];
2813 dateandtimezoneepoch(datenow
, now
);
2816 for (seconds
=checktime
; seconds
<3600L*24*90; seconds
+=checktime
) {
2817 // Check for each device whether a test will be run
2818 time_t testtime
= now
+ seconds
;
2819 for (unsigned i
= 0; i
< numdev
; i
++) {
2820 const dev_config
& cfg
= configs
.at(i
);
2821 dev_state
& state
= states
.at(i
);
2823 char testtype
= next_scheduled_test(cfg
, state
, devices
.at(i
)->is_scsi(), testtime
);
2824 if (testtype
&& (p
= strchr(test_type_chars
, testtype
))) {
2825 unsigned t
= (p
- test_type_chars
);
2826 // Report at most 5 tests of each type
2827 if (++testcnts
[i
*num_test_types
+ t
] <= 5) {
2828 dateandtimezoneepoch(date
, testtime
);
2829 PrintOut(LOG_INFO
, "Device: %s, will do test %d of type %c at %s\n", cfg
.name
.c_str(),
2830 testcnts
[i
*num_test_types
+ t
], testtype
, date
);
2837 dateandtimezoneepoch(date
, now
+seconds
);
2838 PrintOut(LOG_INFO
, "\nTotals [%s - %s]:\n", datenow
, date
);
2839 for (unsigned i
= 0; i
< numdev
; i
++) {
2840 const dev_config
& cfg
= configs
.at(i
);
2841 bool scsi
= devices
.at(i
)->is_scsi();
2842 for (unsigned t
= 0; t
< num_test_types
; t
++) {
2843 int cnt
= testcnts
[i
*num_test_types
+ t
];
2844 if (cnt
== 0 && !strchr((scsi
? "LS" : "LSCO"), test_type_chars
[t
]))
2846 PrintOut(LOG_INFO
, "Device: %s, will do %3d test%s of type %c\n", cfg
.name
.c_str(),
2847 cnt
, (cnt
==1?"":"s"), test_type_chars
[t
]);
2853 // Return zero on success, nonzero on failure. Perform offline (background)
2854 // short or long (extended) self test on given scsi device.
2855 static int DoSCSISelfTest(const dev_config
& cfg
, dev_state
& state
, scsi_device
* device
, char testtype
)
2858 const char *testname
= 0;
2859 const char *name
= cfg
.name
.c_str();
2862 if (scsiSelfTestInProgress(device
, &inProgress
)) {
2863 PrintOut(LOG_CRIT
, "Device: %s, does not support Self-Tests\n", name
);
2864 state
.not_cap_short
= state
.not_cap_long
= true;
2868 if (1 == inProgress
) {
2869 PrintOut(LOG_INFO
, "Device: %s, skip since Self-Test already in "
2870 "progress.\n", name
);
2876 testname
= "Short Self";
2877 retval
= scsiSmartShortSelfTest(device
);
2880 testname
= "Long Self";
2881 retval
= scsiSmartExtendSelfTest(device
);
2884 // If we can't do the test, exit
2885 if (NULL
== testname
) {
2886 PrintOut(LOG_CRIT
, "Device: %s, not capable of %c Self-Test\n", name
,
2891 if ((SIMPLE_ERR_BAD_OPCODE
== retval
) ||
2892 (SIMPLE_ERR_BAD_FIELD
== retval
)) {
2893 PrintOut(LOG_CRIT
, "Device: %s, not capable of %s-Test\n", name
,
2896 state
.not_cap_long
= true;
2898 state
.not_cap_short
= true;
2902 PrintOut(LOG_CRIT
, "Device: %s, execute %s-Test failed (err: %d)\n", name
,
2907 PrintOut(LOG_INFO
, "Device: %s, starting scheduled %s-Test.\n", name
, testname
);
2912 // Do an offline immediate or self-test. Return zero on success,
2913 // nonzero on failure.
2914 static int DoATASelfTest(const dev_config
& cfg
, dev_state
& state
, ata_device
* device
, char testtype
)
2916 const char *name
= cfg
.name
.c_str();
2918 // Read current smart data and check status/capability
2919 struct ata_smart_values data
;
2920 if (ataReadSmartValues(device
, &data
) || !(data
.offline_data_collection_capability
)) {
2921 PrintOut(LOG_CRIT
, "Device: %s, not capable of Offline or Self-Testing.\n", name
);
2925 // Check for capability to do the test
2926 int dotest
= -1, mode
= 0;
2927 const char *testname
= 0;
2930 testname
="Offline Immediate ";
2931 if (isSupportExecuteOfflineImmediate(&data
))
2932 dotest
=OFFLINE_FULL_SCAN
;
2934 state
.not_cap_offline
= true;
2937 testname
="Conveyance Self-";
2938 if (isSupportConveyanceSelfTest(&data
))
2939 dotest
=CONVEYANCE_SELF_TEST
;
2941 state
.not_cap_conveyance
= true;
2944 testname
="Short Self-";
2945 if (isSupportSelfTest(&data
))
2946 dotest
=SHORT_SELF_TEST
;
2948 state
.not_cap_short
= true;
2951 testname
="Long Self-";
2952 if (isSupportSelfTest(&data
))
2953 dotest
=EXTEND_SELF_TEST
;
2955 state
.not_cap_long
= true;
2958 case 'c': case 'n': case 'r':
2959 testname
= "Selective Self-";
2960 if (isSupportSelectiveSelfTest(&data
)) {
2961 dotest
= SELECTIVE_SELF_TEST
;
2963 case 'c': mode
= SEL_CONT
; break;
2964 case 'n': mode
= SEL_NEXT
; break;
2965 case 'r': mode
= SEL_REDO
; break;
2969 state
.not_cap_selective
= true;
2973 // If we can't do the test, exit
2975 PrintOut(LOG_CRIT
, "Device: %s, not capable of %sTest\n", name
, testname
);
2979 // If currently running a self-test, do not interrupt it to start another.
2980 if (15==(data
.self_test_exec_status
>> 4)) {
2981 if (cfg
.firmwarebugs
.is_set(BUG_SAMSUNG3
) && data
.self_test_exec_status
== 0xf0) {
2982 PrintOut(LOG_INFO
, "Device: %s, will not skip scheduled %sTest "
2983 "despite unclear Self-Test byte (SAMSUNG Firmware bug).\n", name
, testname
);
2985 PrintOut(LOG_INFO
, "Device: %s, skip scheduled %sTest; %1d0%% remaining of current Self-Test.\n",
2986 name
, testname
, (int)(data
.self_test_exec_status
& 0x0f));
2991 if (dotest
== SELECTIVE_SELF_TEST
) {
2993 ata_selective_selftest_args selargs
, prev_args
;
2994 selargs
.num_spans
= 1;
2995 selargs
.span
[0].mode
= mode
;
2996 prev_args
.num_spans
= 1;
2997 prev_args
.span
[0].start
= state
.selective_test_last_start
;
2998 prev_args
.span
[0].end
= state
.selective_test_last_end
;
2999 if (ataWriteSelectiveSelfTestLog(device
, selargs
, &data
, state
.num_sectors
, &prev_args
)) {
3000 PrintOut(LOG_CRIT
, "Device: %s, prepare %sTest failed\n", name
, testname
);
3003 uint64_t start
= selargs
.span
[0].start
, end
= selargs
.span
[0].end
;
3004 PrintOut(LOG_INFO
, "Device: %s, %s test span at LBA %" PRIu64
" - %" PRIu64
" (%" PRIu64
" sectors, %u%% - %u%% of disk).\n",
3005 name
, (selargs
.span
[0].mode
== SEL_NEXT
? "next" : "redo"),
3006 start
, end
, end
- start
+ 1,
3007 (unsigned)((100 * start
+ state
.num_sectors
/2) / state
.num_sectors
),
3008 (unsigned)((100 * end
+ state
.num_sectors
/2) / state
.num_sectors
));
3009 state
.selective_test_last_start
= start
;
3010 state
.selective_test_last_end
= end
;
3013 // execute the test, and return status
3014 int retval
= smartcommandhandler(device
, IMMEDIATE_OFFLINE
, dotest
, NULL
);
3016 PrintOut(LOG_CRIT
, "Device: %s, execute %sTest failed.\n", name
, testname
);
3020 // Report recent test start to do_disable_standby_check()
3021 // and force log of next test status
3022 if (testtype
== 'O')
3023 state
.offline_started
= true;
3025 state
.selftest_started
= true;
3027 PrintOut(LOG_INFO
, "Device: %s, starting scheduled %sTest.\n", name
, testname
);
3031 // Check pending sector count attribute values (-C, -U directives).
3032 static void check_pending(const dev_config
& cfg
, dev_state
& state
,
3033 unsigned char id
, bool increase_only
,
3034 const ata_smart_values
& smartval
,
3035 int mailtype
, const char * msg
)
3037 // Find attribute index
3038 int i
= ata_find_attr_index(id
, smartval
);
3039 if (!(i
>= 0 && ata_find_attr_index(id
, state
.smartval
) == i
))
3042 // No report if no sectors pending.
3043 uint64_t rawval
= ata_get_attr_raw_value(smartval
.vendor_attributes
[i
], cfg
.attribute_defs
);
3045 reset_warning_mail(cfg
, state
, mailtype
, "No more %s", msg
);
3049 // If attribute is not reset, report only sector count increases.
3050 uint64_t prev_rawval
= ata_get_attr_raw_value(state
.smartval
.vendor_attributes
[i
], cfg
.attribute_defs
);
3051 if (!(!increase_only
|| prev_rawval
< rawval
))
3055 std::string s
= strprintf("Device: %s, %" PRId64
" %s", cfg
.name
.c_str(), rawval
, msg
);
3056 if (prev_rawval
> 0 && rawval
!= prev_rawval
)
3057 s
+= strprintf(" (changed %+" PRId64
")", rawval
- prev_rawval
);
3059 PrintOut(LOG_CRIT
, "%s\n", s
.c_str());
3060 MailWarning(cfg
, state
, mailtype
, "%s", s
.c_str());
3061 state
.must_write
= true;
3064 // Format Temperature value
3065 static const char * fmt_temp(unsigned char x
, char (& buf
)[20])
3069 snprintf(buf
, sizeof(buf
), "%u", x
);
3073 // Check Temperature limits
3074 static void CheckTemperature(const dev_config
& cfg
, dev_state
& state
, unsigned char currtemp
, unsigned char triptemp
)
3076 if (!(0 < currtemp
&& currtemp
< 255)) {
3077 PrintOut(LOG_INFO
, "Device: %s, failed to read Temperature\n", cfg
.name
.c_str());
3081 // Update Max Temperature
3082 const char * minchg
= "", * maxchg
= "";
3083 if (currtemp
> state
.tempmax
) {
3086 state
.tempmax
= currtemp
;
3087 state
.must_write
= true;
3091 if (!state
.temperature
) {
3093 if (!state
.tempmin
|| currtemp
< state
.tempmin
)
3094 // Delay Min Temperature update by ~ 30 minutes.
3095 state
.tempmin_delay
= time(0) + CHECKTIME
- 60;
3096 PrintOut(LOG_INFO
, "Device: %s, initial Temperature is %d Celsius (Min/Max %s/%u%s)\n",
3097 cfg
.name
.c_str(), (int)currtemp
, fmt_temp(state
.tempmin
, buf
), state
.tempmax
, maxchg
);
3099 PrintOut(LOG_INFO
, " [trip Temperature is %d Celsius]\n", (int)triptemp
);
3100 state
.temperature
= currtemp
;
3103 if (state
.tempmin_delay
) {
3104 // End Min Temperature update delay if ...
3105 if ( (state
.tempmin
&& currtemp
> state
.tempmin
) // current temp exceeds recorded min,
3106 || (state
.tempmin_delay
<= time(0))) { // or delay time is over.
3107 state
.tempmin_delay
= 0;
3109 state
.tempmin
= 255;
3113 // Update Min Temperature
3114 if (!state
.tempmin_delay
&& currtemp
< state
.tempmin
) {
3115 state
.tempmin
= currtemp
;
3116 state
.must_write
= true;
3117 if (currtemp
!= state
.temperature
)
3122 if (cfg
.tempdiff
&& (*minchg
|| *maxchg
|| abs((int)currtemp
- (int)state
.temperature
) >= cfg
.tempdiff
)) {
3123 PrintOut(LOG_INFO
, "Device: %s, Temperature changed %+d Celsius to %u Celsius (Min/Max %s%s/%u%s)\n",
3124 cfg
.name
.c_str(), (int)currtemp
-(int)state
.temperature
, currtemp
, fmt_temp(state
.tempmin
, buf
), minchg
, state
.tempmax
, maxchg
);
3125 state
.temperature
= currtemp
;
3130 if (cfg
.tempcrit
&& currtemp
>= cfg
.tempcrit
) {
3131 PrintOut(LOG_CRIT
, "Device: %s, Temperature %u Celsius reached critical limit of %u Celsius (Min/Max %s%s/%u%s)\n",
3132 cfg
.name
.c_str(), currtemp
, cfg
.tempcrit
, fmt_temp(state
.tempmin
, buf
), minchg
, state
.tempmax
, maxchg
);
3133 MailWarning(cfg
, state
, 12, "Device: %s, Temperature %d Celsius reached critical limit of %u Celsius (Min/Max %s%s/%u%s)",
3134 cfg
.name
.c_str(), currtemp
, cfg
.tempcrit
, fmt_temp(state
.tempmin
, buf
), minchg
, state
.tempmax
, maxchg
);
3136 else if (cfg
.tempinfo
&& currtemp
>= cfg
.tempinfo
) {
3137 PrintOut(LOG_INFO
, "Device: %s, Temperature %u Celsius reached limit of %u Celsius (Min/Max %s%s/%u%s)\n",
3138 cfg
.name
.c_str(), currtemp
, cfg
.tempinfo
, fmt_temp(state
.tempmin
, buf
), minchg
, state
.tempmax
, maxchg
);
3140 else if (cfg
.tempcrit
) {
3141 unsigned char limit
= (cfg
.tempinfo
? cfg
.tempinfo
: cfg
.tempcrit
-5);
3142 if (currtemp
< limit
)
3143 reset_warning_mail(cfg
, state
, 12, "Temperature %u Celsius dropped below %u Celsius", currtemp
, limit
);
3147 // Check normalized and raw attribute values.
3148 static void check_attribute(const dev_config
& cfg
, dev_state
& state
,
3149 const ata_smart_attribute
& attr
,
3150 const ata_smart_attribute
& prev
,
3152 const ata_smart_threshold_entry
* thresholds
)
3154 // Check attribute and threshold
3155 ata_attr_state attrstate
= ata_get_attr_state(attr
, attridx
, thresholds
, cfg
.attribute_defs
);
3156 if (attrstate
== ATTRSTATE_NON_EXISTING
)
3159 // If requested, check for usage attributes that have failed.
3160 if ( cfg
.usagefailed
&& attrstate
== ATTRSTATE_FAILED_NOW
3161 && !cfg
.monitor_attr_flags
.is_set(attr
.id
, MONITOR_IGN_FAILUSE
)) {
3162 std::string attrname
= ata_get_smart_attr_name(attr
.id
, cfg
.attribute_defs
, cfg
.dev_rpm
);
3163 PrintOut(LOG_CRIT
, "Device: %s, Failed SMART usage Attribute: %d %s.\n", cfg
.name
.c_str(), attr
.id
, attrname
.c_str());
3164 MailWarning(cfg
, state
, 2, "Device: %s, Failed SMART usage Attribute: %d %s.", cfg
.name
.c_str(), attr
.id
, attrname
.c_str());
3165 state
.must_write
= true;
3168 // Return if we're not tracking this type of attribute
3169 bool prefail
= !!ATTRIBUTE_FLAGS_PREFAILURE(attr
.flags
);
3170 if (!( ( prefail
&& cfg
.prefail
)
3171 || (!prefail
&& cfg
.usage
)))
3174 // Return if '-I ID' was specified
3175 if (cfg
.monitor_attr_flags
.is_set(attr
.id
, MONITOR_IGNORE
))
3178 // Issue warning if they don't have the same ID in all structures.
3179 if (attr
.id
!= prev
.id
) {
3180 PrintOut(LOG_INFO
,"Device: %s, same Attribute has different ID numbers: %d = %d\n",
3181 cfg
.name
.c_str(), attr
.id
, prev
.id
);
3185 // Compare normalized values if valid.
3186 bool valchanged
= false;
3187 if (attrstate
> ATTRSTATE_NO_NORMVAL
) {
3188 if (attr
.current
!= prev
.current
)
3192 // Compare raw values if requested.
3193 bool rawchanged
= false;
3194 if (cfg
.monitor_attr_flags
.is_set(attr
.id
, MONITOR_RAW
)) {
3195 if ( ata_get_attr_raw_value(attr
, cfg
.attribute_defs
)
3196 != ata_get_attr_raw_value(prev
, cfg
.attribute_defs
))
3200 // Return if no change
3201 if (!(valchanged
|| rawchanged
))
3204 // Format value strings
3205 std::string currstr
, prevstr
;
3206 if (attrstate
== ATTRSTATE_NO_NORMVAL
) {
3207 // Print raw values only
3208 currstr
= strprintf("%s (Raw)",
3209 ata_format_attr_raw_value(attr
, cfg
.attribute_defs
).c_str());
3210 prevstr
= strprintf("%s (Raw)",
3211 ata_format_attr_raw_value(prev
, cfg
.attribute_defs
).c_str());
3213 else if (cfg
.monitor_attr_flags
.is_set(attr
.id
, MONITOR_RAW_PRINT
)) {
3214 // Print normalized and raw values
3215 currstr
= strprintf("%d [Raw %s]", attr
.current
,
3216 ata_format_attr_raw_value(attr
, cfg
.attribute_defs
).c_str());
3217 prevstr
= strprintf("%d [Raw %s]", prev
.current
,
3218 ata_format_attr_raw_value(prev
, cfg
.attribute_defs
).c_str());
3221 // Print normalized values only
3222 currstr
= strprintf("%d", attr
.current
);
3223 prevstr
= strprintf("%d", prev
.current
);
3227 std::string msg
= strprintf("Device: %s, SMART %s Attribute: %d %s changed from %s to %s",
3228 cfg
.name
.c_str(), (prefail
? "Prefailure" : "Usage"), attr
.id
,
3229 ata_get_smart_attr_name(attr
.id
, cfg
.attribute_defs
, cfg
.dev_rpm
).c_str(),
3230 prevstr
.c_str(), currstr
.c_str());
3232 // Report this change as critical ?
3233 if ( (valchanged
&& cfg
.monitor_attr_flags
.is_set(attr
.id
, MONITOR_AS_CRIT
))
3234 || (rawchanged
&& cfg
.monitor_attr_flags
.is_set(attr
.id
, MONITOR_RAW_AS_CRIT
))) {
3235 PrintOut(LOG_CRIT
, "%s\n", msg
.c_str());
3236 MailWarning(cfg
, state
, 2, "%s", msg
.c_str());
3239 PrintOut(LOG_INFO
, "%s\n", msg
.c_str());
3241 state
.must_write
= true;
3245 static int ATACheckDevice(const dev_config
& cfg
, dev_state
& state
, ata_device
* atadev
,
3246 bool firstpass
, bool allow_selftests
)
3248 if (!open_device(cfg
, state
, atadev
, "ATA"))
3251 const char * name
= cfg
.name
.c_str();
3253 // user may have requested (with the -n Directive) to leave the disk
3254 // alone if it is in idle or sleeping mode. In this case check the
3255 // power mode and exit without check if needed
3256 if (cfg
.powermode
&& !state
.powermodefail
) {
3257 int dontcheck
=0, powermode
=ataCheckPowerMode(atadev
);
3258 const char * mode
= 0;
3259 if (0 <= powermode
&& powermode
< 0xff) {
3260 // wait for possible spin up and check again
3263 powermode2
= ataCheckPowerMode(atadev
);
3264 if (powermode2
> powermode
)
3265 PrintOut(LOG_INFO
, "Device: %s, CHECK POWER STATUS spins up disk (0x%02x -> 0x%02x)\n", name
, powermode
, powermode2
);
3266 powermode
= powermode2
;
3273 if (cfg
.powermode
>=1)
3279 if (cfg
.powermode
>=2)
3285 if (cfg
.powermode
>=2)
3291 if (cfg
.powermode
>=3)
3297 if (cfg
.powermode
>=3)
3303 if (cfg
.powermode
>=3)
3309 if (cfg
.powermode
>=3)
3318 mode
="ACTIVE or IDLE";
3322 PrintOut(LOG_CRIT
, "Device: %s, CHECK POWER STATUS returned %d, not ATA compliant, ignoring -n Directive\n",
3324 state
.powermodefail
= true;
3328 // if we are going to skip a check, return now
3330 // skip at most powerskipmax checks
3331 if (!cfg
.powerskipmax
|| state
.powerskipcnt
<cfg
.powerskipmax
) {
3332 CloseDevice(atadev
, name
);
3333 // report first only except if state has changed, avoid waking up system disk
3334 if ((!state
.powerskipcnt
|| state
.lastpowermodeskipped
!= powermode
) && !cfg
.powerquiet
) {
3335 PrintOut(LOG_INFO
, "Device: %s, is in %s mode, suspending checks\n", name
, mode
);
3336 state
.lastpowermodeskipped
= powermode
;
3338 state
.powerskipcnt
++;
3342 PrintOut(LOG_INFO
, "Device: %s, %s mode ignored due to reached limit of skipped checks (%d check%s skipped)\n",
3343 name
, mode
, state
.powerskipcnt
, (state
.powerskipcnt
==1?"":"s"));
3345 state
.powerskipcnt
= 0;
3346 state
.tempmin_delay
= time(0) + CHECKTIME
- 60; // Delay Min Temperature update
3348 else if (state
.powerskipcnt
) {
3349 PrintOut(LOG_INFO
, "Device: %s, is back in %s mode, resuming checks (%d check%s skipped)\n",
3350 name
, mode
, state
.powerskipcnt
, (state
.powerskipcnt
==1?"":"s"));
3351 state
.powerskipcnt
= 0;
3352 state
.tempmin_delay
= time(0) + CHECKTIME
- 60; // Delay Min Temperature update
3356 // check smart status
3357 if (cfg
.smartcheck
) {
3358 int status
=ataSmartStatus2(atadev
);
3360 PrintOut(LOG_INFO
,"Device: %s, not capable of SMART self-check\n",name
);
3361 MailWarning(cfg
, state
, 5, "Device: %s, not capable of SMART self-check", name
);
3362 state
.must_write
= true;
3364 else if (status
==1){
3365 PrintOut(LOG_CRIT
, "Device: %s, FAILED SMART self-check. BACK UP DATA NOW!\n", name
);
3366 MailWarning(cfg
, state
, 1, "Device: %s, FAILED SMART self-check. BACK UP DATA NOW!", name
);
3367 state
.must_write
= true;
3371 // Check everything that depends upon SMART Data (eg, Attribute values)
3372 if ( cfg
.usagefailed
|| cfg
.prefail
|| cfg
.usage
3373 || cfg
.curr_pending_id
|| cfg
.offl_pending_id
3374 || cfg
.tempdiff
|| cfg
.tempinfo
|| cfg
.tempcrit
3375 || cfg
.selftest
|| cfg
.offlinests
|| cfg
.selfteststs
) {
3377 // Read current attribute values.
3378 ata_smart_values curval
;
3379 if (ataReadSmartValues(atadev
, &curval
)){
3380 PrintOut(LOG_CRIT
, "Device: %s, failed to read SMART Attribute Data\n", name
);
3381 MailWarning(cfg
, state
, 6, "Device: %s, failed to read SMART Attribute Data", name
);
3382 state
.must_write
= true;
3385 reset_warning_mail(cfg
, state
, 6, "read SMART Attribute Data worked again");
3387 // look for current or offline pending sectors
3388 if (cfg
.curr_pending_id
)
3389 check_pending(cfg
, state
, cfg
.curr_pending_id
, cfg
.curr_pending_incr
, curval
, 10,
3390 (!cfg
.curr_pending_incr
? "Currently unreadable (pending) sectors"
3391 : "Total unreadable (pending) sectors" ));
3393 if (cfg
.offl_pending_id
)
3394 check_pending(cfg
, state
, cfg
.offl_pending_id
, cfg
.offl_pending_incr
, curval
, 11,
3395 (!cfg
.offl_pending_incr
? "Offline uncorrectable sectors"
3396 : "Total offline uncorrectable sectors"));
3398 // check temperature limits
3399 if (cfg
.tempdiff
|| cfg
.tempinfo
|| cfg
.tempcrit
)
3400 CheckTemperature(cfg
, state
, ata_return_temperature_value(&curval
, cfg
.attribute_defs
), 0);
3402 // look for failed usage attributes, or track usage or prefail attributes
3403 if (cfg
.usagefailed
|| cfg
.prefail
|| cfg
.usage
) {
3404 for (int i
= 0; i
< NUMBER_ATA_SMART_ATTRIBUTES
; i
++) {
3405 check_attribute(cfg
, state
,
3406 curval
.vendor_attributes
[i
],
3407 state
.smartval
.vendor_attributes
[i
],
3408 i
, state
.smartthres
.thres_entries
);
3412 // Log changes of offline data collection status
3413 if (cfg
.offlinests
) {
3414 if ( curval
.offline_data_collection_status
3415 != state
.smartval
.offline_data_collection_status
3416 || state
.offline_started
// test was started in previous call
3417 || (firstpass
&& (debugmode
|| (curval
.offline_data_collection_status
& 0x7d))))
3418 log_offline_data_coll_status(name
, curval
.offline_data_collection_status
);
3421 // Log changes of self-test execution status
3422 if (cfg
.selfteststs
) {
3423 if ( curval
.self_test_exec_status
!= state
.smartval
.self_test_exec_status
3424 || state
.selftest_started
// test was started in previous call
3425 || (firstpass
&& (debugmode
|| curval
.self_test_exec_status
!= 0x00)))
3426 log_self_test_exec_status(name
, curval
.self_test_exec_status
);
3429 // Save the new values for the next time around
3430 state
.smartval
= curval
;
3433 state
.offline_started
= state
.selftest_started
= false;
3435 // check if number of selftest errors has increased (note: may also DECREASE)
3437 CheckSelfTestLogs(cfg
, state
, SelfTestErrorCount(atadev
, name
, cfg
.firmwarebugs
));
3439 // check if number of ATA errors has increased
3440 if (cfg
.errorlog
|| cfg
.xerrorlog
) {
3442 int errcnt1
= -1, errcnt2
= -1;
3444 errcnt1
= read_ata_error_count(atadev
, name
, cfg
.firmwarebugs
, false);
3446 errcnt2
= read_ata_error_count(atadev
, name
, cfg
.firmwarebugs
, true);
3448 // new number of errors is max of both logs
3449 int newc
= (errcnt1
>= errcnt2
? errcnt1
: errcnt2
);
3451 // did command fail?
3453 // lack of PrintOut here is INTENTIONAL
3454 MailWarning(cfg
, state
, 7, "Device: %s, Read SMART Error Log Failed", name
);
3456 // has error count increased?
3457 int oldc
= state
.ataerrorcount
;
3459 PrintOut(LOG_CRIT
, "Device: %s, ATA error count increased from %d to %d\n",
3461 MailWarning(cfg
, state
, 4, "Device: %s, ATA error count increased from %d to %d",
3463 state
.must_write
= true;
3467 state
.ataerrorcount
=newc
;
3470 // if the user has asked, and device is capable (or we're not yet
3471 // sure) check whether a self test should be done now.
3472 if (allow_selftests
&& !cfg
.test_regex
.empty()) {
3473 char testtype
= next_scheduled_test(cfg
, state
, false/*!scsi*/);
3475 DoATASelfTest(cfg
, state
, atadev
, testtype
);
3478 // Don't leave device open -- the OS/user may want to access it
3479 // before the next smartd cycle!
3480 CloseDevice(atadev
, name
);
3482 // Copy ATA attribute values to persistent state
3483 state
.update_persistent_state();
3488 static int SCSICheckDevice(const dev_config
& cfg
, dev_state
& state
, scsi_device
* scsidev
, bool allow_selftests
)
3490 if (!open_device(cfg
, state
, scsidev
, "SCSI"))
3493 const char * name
= cfg
.name
.c_str();
3495 UINT8 asc
= 0, ascq
= 0;
3496 UINT8 currenttemp
= 0, triptemp
= 0;
3497 if (!state
.SuppressReport
) {
3498 if (scsiCheckIE(scsidev
, state
.SmartPageSupported
, state
.TempPageSupported
,
3499 &asc
, &ascq
, ¤ttemp
, &triptemp
)) {
3500 PrintOut(LOG_INFO
, "Device: %s, failed to read SMART values\n",
3502 MailWarning(cfg
, state
, 6, "Device: %s, failed to read SMART values", name
);
3503 state
.SuppressReport
= 1;
3507 const char * cp
= scsiGetIEString(asc
, ascq
);
3509 PrintOut(LOG_CRIT
, "Device: %s, SMART Failure: %s\n", name
, cp
);
3510 MailWarning(cfg
, state
, 1,"Device: %s, SMART Failure: %s", name
, cp
);
3511 } else if (asc
== 4 && ascq
== 9) {
3512 PrintOut(LOG_INFO
,"Device: %s, self-test in progress\n", name
);
3513 } else if (debugmode
)
3514 PrintOut(LOG_INFO
,"Device: %s, non-SMART asc,ascq: %d,%d\n",
3515 name
, (int)asc
, (int)ascq
);
3516 } else if (debugmode
)
3517 PrintOut(LOG_INFO
,"Device: %s, SMART health: passed\n", name
);
3519 // check temperature limits
3520 if (cfg
.tempdiff
|| cfg
.tempinfo
|| cfg
.tempcrit
)
3521 CheckTemperature(cfg
, state
, currenttemp
, triptemp
);
3523 // check if number of selftest errors has increased (note: may also DECREASE)
3525 CheckSelfTestLogs(cfg
, state
, scsiCountFailedSelfTests(scsidev
, 0));
3527 if (allow_selftests
&& !cfg
.test_regex
.empty()) {
3528 char testtype
= next_scheduled_test(cfg
, state
, true/*scsi*/);
3530 DoSCSISelfTest(cfg
, state
, scsidev
, testtype
);
3532 if (!cfg
.attrlog_file
.empty()){
3533 // saving error counters to state
3535 if (state
.ReadECounterPageSupported
&& (0 == scsiLogSense(scsidev
,
3536 READ_ERROR_COUNTER_LPAGE
, 0, tBuf
, sizeof(tBuf
), 0))) {
3537 scsiDecodeErrCounterPage(tBuf
, &state
.scsi_error_counters
[0].errCounter
);
3538 state
.scsi_error_counters
[0].found
=1;
3540 if (state
.WriteECounterPageSupported
&& (0 == scsiLogSense(scsidev
,
3541 WRITE_ERROR_COUNTER_LPAGE
, 0, tBuf
, sizeof(tBuf
), 0))) {
3542 scsiDecodeErrCounterPage(tBuf
, &state
.scsi_error_counters
[1].errCounter
);
3543 state
.scsi_error_counters
[1].found
=1;
3545 if (state
.VerifyECounterPageSupported
&& (0 == scsiLogSense(scsidev
,
3546 VERIFY_ERROR_COUNTER_LPAGE
, 0, tBuf
, sizeof(tBuf
), 0))) {
3547 scsiDecodeErrCounterPage(tBuf
, &state
.scsi_error_counters
[2].errCounter
);
3548 state
.scsi_error_counters
[2].found
=1;
3550 if (state
.NonMediumErrorPageSupported
&& (0 == scsiLogSense(scsidev
,
3551 NON_MEDIUM_ERROR_LPAGE
, 0, tBuf
, sizeof(tBuf
), 0))) {
3552 scsiDecodeNonMediumErrPage(tBuf
, &state
.scsi_nonmedium_error
.nme
);
3553 state
.scsi_nonmedium_error
.found
=1;
3555 // store temperature if not done by CheckTemperature() above
3556 if (!(cfg
.tempdiff
|| cfg
.tempinfo
|| cfg
.tempcrit
))
3557 state
.temperature
= currenttemp
;
3559 CloseDevice(scsidev
, name
);
3563 static int NVMeCheckDevice(const dev_config
& cfg
, dev_state
& state
, nvme_device
* nvmedev
)
3565 if (!open_device(cfg
, state
, nvmedev
, "NVMe"))
3568 const char * name
= cfg
.name
.c_str();
3570 // Read SMART/Health log
3571 nvme_smart_log smart_log
;
3572 if (!nvme_read_smart_log(nvmedev
, smart_log
)) {
3573 PrintOut(LOG_INFO
, "Device: %s, failed to read NVMe SMART/Health Information\n", name
);
3574 MailWarning(cfg
, state
, 6, "Device: %s, failed to read NVMe SMART/Health Information", name
);
3575 state
.must_write
= true;
3579 // Check Critical Warning bits
3580 if (cfg
.smartcheck
&& smart_log
.critical_warning
) {
3581 unsigned char w
= smart_log
.critical_warning
;
3583 static const char * const wnames
[] =
3584 {"LowSpare", "Temperature", "Reliability", "R/O", "VolMemBackup"};
3586 for (unsigned b
= 0, cnt
= 0; b
< 8 ; b
++) {
3587 if (!(w
& (1 << b
)))
3592 msg
+= "..."; break;
3594 if (b
>= sizeof(wnames
)/sizeof(wnames
[0])) {
3595 msg
+= "*Unknown*"; break;
3600 PrintOut(LOG_CRIT
, "Device: %s, Critical Warning (0x%02x): %s\n", name
, w
, msg
.c_str());
3601 MailWarning(cfg
, state
, 1, "Device: %s, Critical Warning (0x%02x): %s", name
, w
, msg
.c_str());
3602 state
.must_write
= true;
3605 // Check temperature limits
3606 if (cfg
.tempdiff
|| cfg
.tempinfo
|| cfg
.tempcrit
) {
3607 int k
= nvme_get_max_temp_kelvin(smart_log
);
3608 // Convert Kelvin to positive Celsius (TODO: Allow negative temperatures)
3614 CheckTemperature(cfg
, state
, c
, 0);
3617 // Check if number of errors has increased
3618 if (cfg
.errorlog
|| cfg
.xerrorlog
) {
3619 uint64_t oldcnt
= state
.nvme_err_log_entries
;
3620 uint64_t newcnt
= le128_to_uint64(smart_log
.num_err_log_entries
);
3621 if (newcnt
> oldcnt
) {
3622 PrintOut(LOG_CRIT
, "Device: %s, number of Error Log entries increased from %" PRIu64
" to %" PRIu64
"\n",
3623 name
, oldcnt
, newcnt
);
3624 MailWarning(cfg
, state
, 4, "Device: %s, number of Error Log entries increased from %" PRIu64
" to %" PRIu64
,
3625 name
, oldcnt
, newcnt
);
3626 state
.must_write
= true;
3628 state
.nvme_err_log_entries
= newcnt
;
3631 CloseDevice(nvmedev
, name
);
3635 // 0=not used, 1=not disabled, 2=disable rejected by OS, 3=disabled
3636 static int standby_disable_state
= 0;
3638 static void init_disable_standby_check(dev_config_vector
& configs
)
3640 // Check for '-l offlinests,ns' or '-l selfteststs,ns' directives
3641 bool sts1
= false, sts2
= false;
3642 for (unsigned i
= 0; i
< configs
.size() && !(sts1
|| sts2
); i
++) {
3643 const dev_config
& cfg
= configs
.at(i
);
3644 if (cfg
.offlinests_ns
)
3646 if (cfg
.selfteststs_ns
)
3650 // Check for support of disable auto standby
3651 // Reenable standby if smartd.conf was reread
3652 if (sts1
|| sts2
|| standby_disable_state
== 3) {
3653 if (!smi()->disable_system_auto_standby(false)) {
3654 if (standby_disable_state
== 3)
3655 PrintOut(LOG_CRIT
, "System auto standby enable failed: %s\n", smi()->get_errmsg());
3657 PrintOut(LOG_INFO
, "Disable auto standby not supported, ignoring ',ns' from %s%s%s\n",
3658 (sts1
? "-l offlinests,ns" : ""), (sts1
&& sts2
? " and " : ""), (sts2
? "-l selfteststs,ns" : ""));
3659 sts1
= sts2
= false;
3664 standby_disable_state
= (sts1
|| sts2
? 1 : 0);
3667 static void do_disable_standby_check(const dev_config_vector
& configs
, const dev_state_vector
& states
)
3669 if (!standby_disable_state
)
3672 // Check for just started or still running self-tests
3673 bool running
= false;
3674 for (unsigned i
= 0; i
< configs
.size() && !running
; i
++) {
3675 const dev_config
& cfg
= configs
.at(i
); const dev_state
& state
= states
.at(i
);
3677 if ( ( cfg
.offlinests_ns
3678 && (state
.offline_started
||
3679 is_offl_coll_in_progress(state
.smartval
.offline_data_collection_status
)))
3680 || ( cfg
.selfteststs_ns
3681 && (state
.selftest_started
||
3682 is_self_test_in_progress(state
.smartval
.self_test_exec_status
))) )
3684 // state.offline/selftest_started will be reset after next logging of test status
3687 // Disable/enable auto standby and log state changes
3689 if (standby_disable_state
!= 1) {
3690 if (!smi()->disable_system_auto_standby(false))
3691 PrintOut(LOG_CRIT
, "Self-test(s) completed, system auto standby enable failed: %s\n",
3692 smi()->get_errmsg());
3694 PrintOut(LOG_INFO
, "Self-test(s) completed, system auto standby enabled\n");
3695 standby_disable_state
= 1;
3698 else if (!smi()->disable_system_auto_standby(true)) {
3699 if (standby_disable_state
!= 2) {
3700 PrintOut(LOG_INFO
, "Self-test(s) in progress, system auto standby disable rejected: %s\n",
3701 smi()->get_errmsg());
3702 standby_disable_state
= 2;
3706 if (standby_disable_state
!= 3) {
3707 PrintOut(LOG_INFO
, "Self-test(s) in progress, system auto standby disabled\n");
3708 standby_disable_state
= 3;
3713 // Checks the SMART status of all ATA and SCSI devices
3714 static void CheckDevicesOnce(const dev_config_vector
& configs
, dev_state_vector
& states
,
3715 smart_device_list
& devices
, bool firstpass
, bool allow_selftests
)
3717 for (unsigned i
= 0; i
< configs
.size(); i
++) {
3718 const dev_config
& cfg
= configs
.at(i
);
3719 dev_state
& state
= states
.at(i
);
3720 smart_device
* dev
= devices
.at(i
);
3722 ATACheckDevice(cfg
, state
, dev
->to_ata(), firstpass
, allow_selftests
);
3723 else if (dev
->is_scsi())
3724 SCSICheckDevice(cfg
, state
, dev
->to_scsi(), allow_selftests
);
3725 else if (dev
->is_nvme())
3726 NVMeCheckDevice(cfg
, state
, dev
->to_nvme());
3729 do_disable_standby_check(configs
, states
);
3732 // Set if Initialize() was called
3733 static bool is_initialized
= false;
3735 // Does initialization right after fork to daemon mode
3736 static void Initialize(time_t *wakeuptime
)
3738 // Call Goodbye() on exit
3739 is_initialized
= true;
3745 // install signal handlers. On Solaris, can't use signal() because
3746 // it resets the handler to SIG_DFL after each call. So use sigset()
3747 // instead. So SIGNALFN()==signal() or SIGNALFN()==sigset().
3749 // normal and abnormal exit
3750 if (SIGNALFN(SIGTERM
, sighandler
)==SIG_IGN
)
3751 SIGNALFN(SIGTERM
, SIG_IGN
);
3752 if (SIGNALFN(SIGQUIT
, sighandler
)==SIG_IGN
)
3753 SIGNALFN(SIGQUIT
, SIG_IGN
);
3755 // in debug mode, <CONTROL-C> ==> HUP
3756 if (SIGNALFN(SIGINT
, debugmode
?HUPhandler
:sighandler
)==SIG_IGN
)
3757 SIGNALFN(SIGINT
, SIG_IGN
);
3759 // Catch HUP and USR1
3760 if (SIGNALFN(SIGHUP
, HUPhandler
)==SIG_IGN
)
3761 SIGNALFN(SIGHUP
, SIG_IGN
);
3762 if (SIGNALFN(SIGUSR1
, USR1handler
)==SIG_IGN
)
3763 SIGNALFN(SIGUSR1
, SIG_IGN
);
3765 if (SIGNALFN(SIGUSR2
, USR2handler
)==SIG_IGN
)
3766 SIGNALFN(SIGUSR2
, SIG_IGN
);
3769 // initialize wakeup time to CURRENT time
3770 *wakeuptime
=time(NULL
);
3776 // Toggle debug mode implemented for native windows only
3777 // (there is no easy way to reopen tty on *nix)
3778 static void ToggleDebugMode()
3781 PrintOut(LOG_INFO
,"Signal USR2 - enabling debug mode\n");
3782 if (!daemon_enable_console("smartd [Debug]")) {
3784 daemon_signal(SIGINT
, HUPhandler
);
3785 PrintOut(LOG_INFO
,"smartd debug mode enabled, PID=%d\n", getpid());
3788 PrintOut(LOG_INFO
,"enable console failed\n");
3790 else if (debugmode
== 1) {
3791 daemon_disable_console();
3793 daemon_signal(SIGINT
, sighandler
);
3794 PrintOut(LOG_INFO
,"Signal USR2 - debug mode disabled\n");
3797 PrintOut(LOG_INFO
,"Signal USR2 - debug mode %d not changed\n", debugmode
);
3801 static time_t dosleep(time_t wakeuptime
, bool & sigwakeup
)
3803 // If past wake-up-time, compute next wake-up-time
3804 time_t timenow
=time(NULL
);
3805 while (wakeuptime
<=timenow
){
3806 int intervals
=1+(timenow
-wakeuptime
)/checktime
;
3807 wakeuptime
+=intervals
*checktime
;
3810 // sleep until we catch SIGUSR1 or have completed sleeping
3812 while (timenow
< wakeuptime
+addtime
&& !caughtsigUSR1
&& !caughtsigHUP
&& !caughtsigEXIT
) {
3814 // protect user again system clock being adjusted backwards
3815 if (wakeuptime
>timenow
+checktime
){
3816 PrintOut(LOG_CRIT
, "System clock time adjusted to the past. Resetting next wakeup time.\n");
3817 wakeuptime
=timenow
+checktime
;
3820 // Exit sleep when time interval has expired or a signal is received
3821 sleep(wakeuptime
+addtime
-timenow
);
3824 // toggle debug mode?
3825 if (caughtsigUSR2
) {
3833 // Actual sleep time too long?
3834 if (!addtime
&& timenow
> wakeuptime
+60) {
3836 PrintOut(LOG_INFO
, "Sleep time was %d seconds too long, assuming wakeup from standby mode.\n",
3837 (int)(timenow
-wakeuptime
));
3838 // Wait another 20 seconds to avoid I/O errors during disk spin-up
3839 addtime
= timenow
-wakeuptime
+20;
3840 // Use next wake-up-time if close
3841 int nextcheck
= checktime
- addtime
% checktime
;
3842 if (nextcheck
<= 20)
3843 addtime
+= nextcheck
;
3847 // if we caught a SIGUSR1 then print message and clear signal
3849 PrintOut(LOG_INFO
,"Signal USR1 - checking devices now rather than in %d seconds.\n",
3850 wakeuptime
-timenow
>0?(int)(wakeuptime
-timenow
):0);
3855 // return adjusted wakeuptime
3859 // Print out a list of valid arguments for the Directive d
3860 static void printoutvaliddirectiveargs(int priority
, char d
)
3864 PrintOut(priority
, "never[,N][,q], sleep[,N][,q], standby[,N][,q], idle[,N][,q]");
3867 PrintOut(priority
, "valid_regular_expression");
3870 PrintOut(priority
, "%s", smi()->get_valid_dev_types_str().c_str());
3873 PrintOut(priority
, "normal, permissive");
3877 PrintOut(priority
, "on, off");
3880 PrintOut(priority
, "error, selftest");
3883 PrintOut(priority
, "\"once\", \"daily\", \"diminishing\", \"test\", \"exec\"");
3886 PrintOut(priority
, "\n%s\n", create_vendor_attribute_arg_list().c_str());
3889 PrintOut(priority
, "use, ignore, show, showall");
3892 PrintOut(priority
, "%s", get_valid_firmwarebug_args());
3895 PrintOut(priority
, "aam,[N|off], apm,[N|off], lookahead,[on|off], dsn,[on|off] "
3896 "security-freeze, standby,[N|off], wcache,[on|off]");
3901 // exits with an error message, or returns integer value of token
3902 static int GetInteger(const char *arg
, const char *name
, const char *token
, int lineno
, const char *cfgfile
,
3903 int min
, int max
, char * suffix
= 0)
3905 // make sure argument is there
3907 PrintOut(LOG_CRIT
,"File %s line %d (drive %s): Directive: %s takes integer argument from %d to %d.\n",
3908 cfgfile
, lineno
, name
, token
, min
, max
);
3912 // get argument value (base 10), check that it's integer, and in-range
3914 int val
= strtol(arg
,&endptr
,10);
3916 // optional suffix present?
3918 if (!strcmp(endptr
, suffix
))
3919 endptr
+= strlen(suffix
);
3924 if (!(!*endptr
&& min
<= val
&& val
<= max
)) {
3925 PrintOut(LOG_CRIT
,"File %s line %d (drive %s): Directive: %s has argument: %s; needs integer from %d to %d.\n",
3926 cfgfile
, lineno
, name
, token
, arg
, min
, max
);
3930 // all is well; return value
3935 // Get 1-3 small integer(s) for '-W' directive
3936 static int Get3Integers(const char *arg
, const char *name
, const char *token
, int lineno
, const char *cfgfile
,
3937 unsigned char *val1
, unsigned char *val2
, unsigned char *val3
)
3939 unsigned v1
= 0, v2
= 0, v3
= 0;
3940 int n1
= -1, n2
= -1, n3
= -1, len
;
3942 PrintOut(LOG_CRIT
,"File %s line %d (drive %s): Directive: %s takes 1-3 integer argument(s) from 0 to 255.\n",
3943 cfgfile
, lineno
, name
, token
);
3948 if (!( sscanf(arg
, "%u%n,%u%n,%u%n", &v1
, &n1
, &v2
, &n2
, &v3
, &n3
) >= 1
3949 && (n1
== len
|| n2
== len
|| n3
== len
) && v1
<= 255 && v2
<= 255 && v3
<= 255)) {
3950 PrintOut(LOG_CRIT
,"File %s line %d (drive %s): Directive: %s has argument: %s; needs 1-3 integer(s) from 0 to 255.\n",
3951 cfgfile
, lineno
, name
, token
, arg
);
3954 *val1
= (unsigned char)v1
; *val2
= (unsigned char)v2
; *val3
= (unsigned char)v3
;
3961 // Concatenate strtok() results if quoted with "..."
3962 static const char * strtok_dequote(const char * delimiters
)
3964 const char * t
= strtok(0, delimiters
);
3965 if (!t
|| t
[0] != '"')
3968 static std::string token
;
3971 t
= strtok(0, delimiters
);
3975 int len
= strlen(t
);
3976 if (t
[len
-1] == '"') {
3977 token
+= std::string(t
, len
-1);
3982 return token
.c_str();
3988 // This function returns 1 if it has correctly parsed one token (and
3989 // any arguments), else zero if no tokens remain. It returns -1 if an
3990 // error was encountered.
3991 static int ParseToken(char * token
, dev_config
& cfg
, smart_devtype_list
& scan_types
)
3994 const char * name
= cfg
.name
.c_str();
3995 int lineno
=cfg
.lineno
;
3996 const char *delim
= " \n\t";
3999 const char *arg
= 0;
4001 // is the rest of the line a comment
4005 // is the token not recognized?
4006 if (*token
!='-' || strlen(token
)!=2) {
4007 PrintOut(LOG_CRIT
,"File %s line %d (drive %s): unknown Directive: %s\n",
4008 configfile
, lineno
, name
, token
);
4009 PrintOut(LOG_CRIT
, "Run smartd -D to print a list of valid Directives.\n");
4013 // token we will be parsing:
4016 // parse the token and swallow its argument
4018 char plus
[] = "+", excl
[] = "!";
4022 // monitor current pending sector count (default 197)
4023 if ((val
= GetInteger(arg
=strtok(NULL
,delim
), name
, token
, lineno
, configfile
, 0, 255, plus
)) < 0)
4025 cfg
.curr_pending_id
= (unsigned char)val
;
4026 cfg
.curr_pending_incr
= (*plus
== '+');
4027 cfg
.curr_pending_set
= true;
4030 // monitor offline uncorrectable sectors (default 198)
4031 if ((val
= GetInteger(arg
=strtok(NULL
,delim
), name
, token
, lineno
, configfile
, 0, 255, plus
)) < 0)
4033 cfg
.offl_pending_id
= (unsigned char)val
;
4034 cfg
.offl_pending_incr
= (*plus
== '+');
4035 cfg
.offl_pending_set
= true;
4038 // Set tolerance level for SMART command failures
4039 if ((arg
= strtok(NULL
, delim
)) == NULL
) {
4041 } else if (!strcmp(arg
, "normal")) {
4042 // Normal mode: exit on failure of a mandatory S.M.A.R.T. command, but
4043 // not on failure of an optional S.M.A.R.T. command.
4044 // This is the default so we don't need to actually do anything here.
4045 cfg
.permissive
= false;
4046 } else if (!strcmp(arg
, "permissive")) {
4047 // Permissive mode; ignore errors from Mandatory SMART commands
4048 cfg
.permissive
= true;
4054 // specify the device type
4055 if ((arg
= strtok(NULL
, delim
)) == NULL
) {
4057 } else if (!strcmp(arg
, "ignore")) {
4059 } else if (!strcmp(arg
, "removable")) {
4060 cfg
.removable
= true;
4061 } else if (!strcmp(arg
, "auto")) {
4066 scan_types
.push_back(arg
);
4071 if (!(arg
= strtok(0, delim
)))
4073 else if (!parse_firmwarebug_def(arg
, cfg
.firmwarebugs
))
4077 // check SMART status
4078 cfg
.smartcheck
= true;
4081 // check for failure of usage attributes
4082 cfg
.usagefailed
= true;
4085 // track changes in all vendor attributes
4090 // track changes in prefail vendor attributes
4094 // track changes in usage vendor attributes
4098 // track changes in SMART logs
4099 if ((arg
= strtok(NULL
, delim
)) == NULL
) {
4101 } else if (!strcmp(arg
, "selftest")) {
4102 // track changes in self-test log
4103 cfg
.selftest
= true;
4104 } else if (!strcmp(arg
, "error")) {
4105 // track changes in ATA error log
4106 cfg
.errorlog
= true;
4107 } else if (!strcmp(arg
, "xerror")) {
4108 // track changes in Extended Comprehensive SMART error log
4109 cfg
.xerrorlog
= true;
4110 } else if (!strcmp(arg
, "offlinests")) {
4111 // track changes in offline data collection status
4112 cfg
.offlinests
= true;
4113 } else if (!strcmp(arg
, "offlinests,ns")) {
4114 // track changes in offline data collection status, disable auto standby
4115 cfg
.offlinests
= cfg
.offlinests_ns
= true;
4116 } else if (!strcmp(arg
, "selfteststs")) {
4117 // track changes in self-test execution status
4118 cfg
.selfteststs
= true;
4119 } else if (!strcmp(arg
, "selfteststs,ns")) {
4120 // track changes in self-test execution status, disable auto standby
4121 cfg
.selfteststs
= cfg
.selfteststs_ns
= true;
4122 } else if (!strncmp(arg
, "scterc,", sizeof("scterc,")-1)) {
4123 // set SCT Error Recovery Control
4124 unsigned rt
= ~0, wt
= ~0; int nc
= -1;
4125 sscanf(arg
,"scterc,%u,%u%n", &rt
, &wt
, &nc
);
4126 if (nc
== (int)strlen(arg
) && rt
<= 999 && wt
<= 999) {
4127 cfg
.sct_erc_set
= true;
4128 cfg
.sct_erc_readtime
= rt
;
4129 cfg
.sct_erc_writetime
= wt
;
4138 // monitor everything
4139 cfg
.smartcheck
= true;
4141 cfg
.usagefailed
= true;
4143 cfg
.selftest
= true;
4144 cfg
.errorlog
= true;
4145 cfg
.selfteststs
= true;
4148 // automatic offline testing enable/disable
4149 if ((arg
= strtok(NULL
, delim
)) == NULL
) {
4151 } else if (!strcmp(arg
, "on")) {
4152 cfg
.autoofflinetest
= 2;
4153 } else if (!strcmp(arg
, "off")) {
4154 cfg
.autoofflinetest
= 1;
4160 // skip disk check if in idle or standby mode
4161 if (!(arg
= strtok(NULL
, delim
)))
4164 char *endptr
= NULL
;
4165 char *next
= strchr(const_cast<char*>(arg
), ',');
4167 cfg
.powerquiet
= false;
4168 cfg
.powerskipmax
= 0;
4170 if (next
!=NULL
) *next
='\0';
4171 if (!strcmp(arg
, "never"))
4173 else if (!strcmp(arg
, "sleep"))
4175 else if (!strcmp(arg
, "standby"))
4177 else if (!strcmp(arg
, "idle"))
4182 // if optional arguments are present
4183 if (!badarg
&& next
!=NULL
) {
4185 cfg
.powerskipmax
= strtol(next
, &endptr
, 10);
4187 cfg
.powerskipmax
= 0;
4189 next
= endptr
+ (*endptr
!= '\0');
4190 if (cfg
.powerskipmax
<= 0)
4193 if (*next
!= '\0') {
4194 if (!strcmp("q", next
))
4195 cfg
.powerquiet
= true;
4204 // automatic attribute autosave enable/disable
4205 if ((arg
= strtok(NULL
, delim
)) == NULL
) {
4207 } else if (!strcmp(arg
, "on")) {
4209 } else if (!strcmp(arg
, "off")) {
4216 // warn user, and delete any previously given -s REGEXP Directives
4217 if (!cfg
.test_regex
.empty()){
4218 PrintOut(LOG_INFO
, "File %s line %d (drive %s): ignoring previous Test Directive -s %s\n",
4219 configfile
, lineno
, name
, cfg
.test_regex
.get_pattern());
4220 cfg
.test_regex
= regular_expression();
4222 // check for missing argument
4223 if (!(arg
= strtok(NULL
, delim
))) {
4228 if (!cfg
.test_regex
.compile(arg
, REG_EXTENDED
)) {
4229 // not a valid regular expression!
4230 PrintOut(LOG_CRIT
, "File %s line %d (drive %s): -s argument \"%s\" is INVALID extended regular expression. %s.\n",
4231 configfile
, lineno
, name
, arg
, cfg
.test_regex
.get_errmsg());
4234 // Do a bit of sanity checking and warn user if we think that
4235 // their regexp is "strange". User probably confused about shell
4236 // glob(3) syntax versus regular expression syntax regexp(7).
4237 if (arg
[(val
= strspn(arg
, "0123456789/.-+*|()?^$[]SLCOcnr"))])
4238 PrintOut(LOG_INFO
, "File %s line %d (drive %s): warning, character %d (%c) looks odd in extended regular expression %s\n",
4239 configfile
, lineno
, name
, val
+1, arg
[val
], arg
);
4243 // send email to address that follows
4244 if (!(arg
= strtok(NULL
,delim
)))
4247 if (!cfg
.emailaddress
.empty())
4248 PrintOut(LOG_INFO
, "File %s line %d (drive %s): ignoring previous Address Directive -m %s\n",
4249 configfile
, lineno
, name
, cfg
.emailaddress
.c_str());
4250 #ifdef _WIN32 // TODO: Remove after smartmontools 6.5
4251 if ( !strcmp(arg
, "msgbox") || !strcmp(arg
, "sysmsgbox")
4252 || str_starts_with(arg
, "msgbox,") || str_starts_with(arg
, "sysmsgbox,")) {
4253 PrintOut(LOG_CRIT
, "File %s line %d (drive %s): -m %s is no longer supported, use -m console[,...] instead\n",
4254 configfile
, lineno
, name
, arg
);
4258 cfg
.emailaddress
= arg
;
4262 // email warning options
4263 if (!(arg
= strtok(NULL
, delim
)))
4265 else if (!strcmp(arg
, "once"))
4267 else if (!strcmp(arg
, "daily"))
4269 else if (!strcmp(arg
, "diminishing"))
4271 else if (!strcmp(arg
, "test"))
4273 else if (!strcmp(arg
, "exec")) {
4274 // Get the next argument (the command line)
4276 // Allow "/path name/with spaces/..." on Windows
4277 arg
= strtok_dequote(delim
);
4278 if (arg
&& arg
[0] == '"') {
4279 PrintOut(LOG_CRIT
, "File %s line %d (drive %s): Directive %s 'exec' argument: missing closing quote\n",
4280 configfile
, lineno
, name
, token
);
4284 arg
= strtok(0, delim
);
4287 PrintOut(LOG_CRIT
, "File %s line %d (drive %s): Directive %s 'exec' argument must be followed by executable path.\n",
4288 configfile
, lineno
, name
, token
);
4291 // Free the last cmd line given if any, and copy new one
4292 if (!cfg
.emailcmdline
.empty())
4293 PrintOut(LOG_INFO
, "File %s line %d (drive %s): ignoring previous mail Directive -M exec %s\n",
4294 configfile
, lineno
, name
, cfg
.emailcmdline
.c_str());
4295 cfg
.emailcmdline
= arg
;
4301 // ignore failure of usage attribute
4302 if ((val
=GetInteger(arg
=strtok(NULL
,delim
), name
, token
, lineno
, configfile
, 1, 255))<0)
4304 cfg
.monitor_attr_flags
.set(val
, MONITOR_IGN_FAILUSE
);
4307 // ignore attribute for tracking purposes
4308 if ((val
=GetInteger(arg
=strtok(NULL
,delim
), name
, token
, lineno
, configfile
, 1, 255))<0)
4310 cfg
.monitor_attr_flags
.set(val
, MONITOR_IGNORE
);
4313 // print raw value when tracking
4314 if ((val
= GetInteger(arg
=strtok(NULL
,delim
), name
, token
, lineno
, configfile
, 1, 255, excl
)) < 0)
4316 cfg
.monitor_attr_flags
.set(val
, MONITOR_RAW_PRINT
);
4317 if (*excl
== '!') // attribute change is critical
4318 cfg
.monitor_attr_flags
.set(val
, MONITOR_AS_CRIT
);
4321 // track changes in raw value (forces printing of raw value)
4322 if ((val
= GetInteger(arg
=strtok(NULL
,delim
), name
, token
, lineno
, configfile
, 1, 255, excl
)) < 0)
4324 cfg
.monitor_attr_flags
.set(val
, MONITOR_RAW_PRINT
|MONITOR_RAW
);
4325 if (*excl
== '!') // raw value change is critical
4326 cfg
.monitor_attr_flags
.set(val
, MONITOR_RAW_AS_CRIT
);
4329 // track Temperature
4330 if (Get3Integers(arg
=strtok(NULL
, delim
), name
, token
, lineno
, configfile
,
4331 &cfg
.tempdiff
, &cfg
.tempinfo
, &cfg
.tempcrit
) < 0)
4335 // non-default vendor-specific attribute meaning
4336 if (!(arg
=strtok(NULL
,delim
))) {
4338 } else if (!parse_attribute_def(arg
, cfg
.attribute_defs
, PRIOR_USER
)) {
4343 // Define use of drive-specific presets.
4344 if (!(arg
= strtok(NULL
, delim
))) {
4346 } else if (!strcmp(arg
, "use")) {
4347 cfg
.ignorepresets
= false;
4348 } else if (!strcmp(arg
, "ignore")) {
4349 cfg
.ignorepresets
= true;
4350 } else if (!strcmp(arg
, "show")) {
4351 cfg
.showpresets
= true;
4352 } else if (!strcmp(arg
, "showall")) {
4360 // Various ATA settings
4361 if (!(arg
= strtok(NULL
, delim
))) {
4365 char arg2
[16+1]; unsigned val
;
4366 int n1
= -1, n2
= -1, n3
= -1, len
= strlen(arg
);
4367 if (sscanf(arg
, "%16[^,=]%n%*[,=]%n%u%n", arg2
, &n1
, &n2
, &val
, &n3
) >= 1
4368 && (n1
== len
|| n2
> 0)) {
4369 bool on
= (n2
> 0 && !strcmp(arg
+n2
, "on"));
4370 bool off
= (n2
> 0 && !strcmp(arg
+n2
, "off"));
4374 if (!strcmp(arg2
, "aam")) {
4377 else if (val
<= 254)
4378 cfg
.set_aam
= val
+ 1;
4382 else if (!strcmp(arg2
, "apm")) {
4385 else if (1 <= val
&& val
<= 254)
4386 cfg
.set_apm
= val
+ 1;
4390 else if (!strcmp(arg2
, "lookahead")) {
4392 cfg
.set_lookahead
= -1;
4394 cfg
.set_lookahead
= 1;
4398 else if (!strcmp(arg
, "security-freeze")) {
4399 cfg
.set_security_freeze
= true;
4401 else if (!strcmp(arg2
, "standby")) {
4403 cfg
.set_standby
= 0 + 1;
4404 else if (val
<= 255)
4405 cfg
.set_standby
= val
+ 1;
4409 else if (!strcmp(arg2
, "wcache")) {
4411 cfg
.set_wcache
= -1;
4417 else if (!strcmp(arg2
, "dsn")) {
4434 // Directive not recognized
4435 PrintOut(LOG_CRIT
,"File %s line %d (drive %s): unknown Directive: %s\n",
4436 configfile
, lineno
, name
, token
);
4441 PrintOut(LOG_CRIT
, "File %s line %d (drive %s): Missing argument to %s Directive\n",
4442 configfile
, lineno
, name
, token
);
4445 PrintOut(LOG_CRIT
, "File %s line %d (drive %s): Invalid argument to %s Directive: %s\n",
4446 configfile
, lineno
, name
, token
, arg
);
4448 if (missingarg
|| badarg
) {
4449 PrintOut(LOG_CRIT
, "Valid arguments to %s Directive are: ", token
);
4450 printoutvaliddirectiveargs(LOG_CRIT
, sym
);
4451 PrintOut(LOG_CRIT
, "\n");
4458 // Scan directive for configuration file
4459 #define SCANDIRECTIVE "DEVICESCAN"
4461 // This is the routine that adds things to the conf_entries list.
4463 // Return values are:
4464 // 1: parsed a normal line
4465 // 0: found DEFAULT setting or comment or blank line
4466 // -1: found SCANDIRECTIVE line
4467 // -2: found an error
4469 // Note: this routine modifies *line from the caller!
4470 static int ParseConfigLine(dev_config_vector
& conf_entries
, dev_config
& default_conf
,
4471 smart_devtype_list
& scan_types
, int lineno
, /*const*/ char * line
)
4473 const char *delim
= " \n\t";
4475 // get first token: device name. If a comment, skip line
4476 const char * name
= strtok(line
, delim
);
4477 if (!name
|| *name
== '#')
4480 // Check device name for DEFAULT or DEVICESCAN
4482 if (!strcmp("DEFAULT", name
)) {
4484 // Restart with empty defaults
4485 default_conf
= dev_config();
4488 retval
= (!strcmp(SCANDIRECTIVE
, name
) ? -1 : 1);
4489 // Init new entry with current defaults
4490 conf_entries
.push_back(default_conf
);
4492 dev_config
& cfg
= (retval
? conf_entries
.back() : default_conf
);
4494 cfg
.name
= name
; // Later replaced by dev->get_info().info_name
4495 cfg
.dev_name
= name
; // If DEVICESCAN later replaced by get->dev_info().dev_name
4496 cfg
.lineno
= lineno
;
4498 // parse tokens one at a time from the file.
4499 while (char * token
= strtok(0, delim
)) {
4500 int rc
= ParseToken(token
, cfg
, scan_types
);
4502 // error found on the line
4509 // PrintOut(LOG_INFO,"Parsed token %s\n",token);
4512 // Check for multiple -d TYPE directives
4513 if (retval
!= -1 && scan_types
.size() > 1) {
4514 PrintOut(LOG_CRIT
, "Drive: %s, invalid multiple -d TYPE Directives on line %d of file %s\n",
4515 cfg
.name
.c_str(), cfg
.lineno
, configfile
);
4519 // Don't perform checks below for DEFAULT entries
4523 // If NO monitoring directives are set, then set all of them.
4524 if (!( cfg
.smartcheck
|| cfg
.selftest
4525 || cfg
.errorlog
|| cfg
.xerrorlog
4526 || cfg
.offlinests
|| cfg
.selfteststs
4527 || cfg
.usagefailed
|| cfg
.prefail
|| cfg
.usage
4528 || cfg
.tempdiff
|| cfg
.tempinfo
|| cfg
.tempcrit
)) {
4530 PrintOut(LOG_INFO
,"Drive: %s, implied '-a' Directive on line %d of file %s\n",
4531 cfg
.name
.c_str(), cfg
.lineno
, configfile
);
4533 cfg
.smartcheck
= true;
4534 cfg
.usagefailed
= true;
4537 cfg
.selftest
= true;
4538 cfg
.errorlog
= true;
4539 cfg
.selfteststs
= true;
4542 // additional sanity check. Has user set -M options without -m?
4543 if (cfg
.emailaddress
.empty() && (!cfg
.emailcmdline
.empty() || cfg
.emailfreq
|| cfg
.emailtest
)){
4544 PrintOut(LOG_CRIT
,"Drive: %s, -M Directive(s) on line %d of file %s need -m ADDRESS Directive\n",
4545 cfg
.name
.c_str(), cfg
.lineno
, configfile
);
4549 // has the user has set <nomailer>?
4550 if (cfg
.emailaddress
== "<nomailer>") {
4551 // check that -M exec is also set
4552 if (cfg
.emailcmdline
.empty()){
4553 PrintOut(LOG_CRIT
,"Drive: %s, -m <nomailer> Directive on line %d of file %s needs -M exec Directive\n",
4554 cfg
.name
.c_str(), cfg
.lineno
, configfile
);
4557 // From here on the sign of <nomailer> is cfg.emailaddress.empty() and !cfg.emailcmdline.empty()
4558 cfg
.emailaddress
.clear();
4564 // Parses a configuration file. Return values are:
4565 // N=>0: found N entries
4566 // -1: syntax error in config file
4567 // -2: config file does not exist
4568 // -3: config file exists but cannot be read
4570 // In the case where the return value is 0, there are three
4572 // Empty configuration file ==> conf_entries.empty()
4573 // No configuration file ==> conf_entries[0].lineno == 0
4574 // SCANDIRECTIVE found ==> conf_entries.back().lineno != 0 (size >= 1)
4575 static int ParseConfigFile(dev_config_vector
& conf_entries
, smart_devtype_list
& scan_types
)
4577 // maximum line length in configuration file
4578 const int MAXLINELEN
= 256;
4579 // maximum length of a continued line in configuration file
4580 const int MAXCONTLINE
= 1023;
4583 // Open config file, if it exists and is not <stdin>
4584 if (!(configfile
== configfile_stdin
)) { // pointer comparison ok here
4585 if (!f
.open(configfile
,"r") && (errno
!=ENOENT
|| !configfile_alt
.empty())) {
4586 // file exists but we can't read it or it should exist due to '-c' option
4587 int ret
= (errno
!=ENOENT
? -3 : -2);
4588 PrintOut(LOG_CRIT
,"%s: Unable to open configuration file %s\n",
4589 strerror(errno
),configfile
);
4593 else // read from stdin ('-c -' option)
4596 // Start with empty defaults
4597 dev_config default_conf
;
4599 // No configuration file found -- use fake one
4602 char fakeconfig
[] = SCANDIRECTIVE
" -a"; // TODO: Remove this hack, build cfg_entry.
4604 if (ParseConfigLine(conf_entries
, default_conf
, scan_types
, 0, fakeconfig
) != -1)
4605 throw std::logic_error("Internal error parsing " SCANDIRECTIVE
);
4610 setmode(fileno(f
), O_TEXT
); // Allow files with \r\n
4613 // configuration file exists
4614 PrintOut(LOG_INFO
,"Opened configuration file %s\n",configfile
);
4616 // parse config file line by line
4617 int lineno
= 1, cont
= 0, contlineno
= 0;
4618 char line
[MAXLINELEN
+2];
4619 char fullline
[MAXCONTLINE
+1];
4622 int len
=0,scandevice
;
4627 // make debugging simpler
4628 memset(line
,0,sizeof(line
));
4631 code
=fgets(line
, MAXLINELEN
+2, f
);
4633 // are we at the end of the file?
4636 scandevice
= ParseConfigLine(conf_entries
, default_conf
, scan_types
, contlineno
, fullline
);
4637 // See if we found a SCANDIRECTIVE directive
4640 // did we find a syntax error
4643 // the final line is part of a continuation line
4649 // input file line number
4652 // See if line is too long
4654 if (len
>MAXLINELEN
){
4656 if (line
[len
-1]=='\n')
4657 warn
="(including newline!) ";
4660 PrintOut(LOG_CRIT
,"Error: line %d of file %s %sis more than MAXLINELEN=%d characters.\n",
4661 (int)contlineno
,configfile
,warn
,(int)MAXLINELEN
);
4665 // Ignore anything after comment symbol
4666 if ((comment
=strchr(line
,'#'))){
4671 // is the total line (made of all continuation lines) too long?
4672 if (cont
+len
>MAXCONTLINE
){
4673 PrintOut(LOG_CRIT
,"Error: continued line %d (actual line %d) of file %s is more than MAXCONTLINE=%d characters.\n",
4674 lineno
, (int)contlineno
, configfile
, (int)MAXCONTLINE
);
4678 // copy string so far into fullline, and increment length
4679 snprintf(fullline
+cont
, sizeof(fullline
)-cont
, "%s" ,line
);
4682 // is this a continuation line. If so, replace \ by space and look at next line
4683 if ( (lastslash
=strrchr(line
,'\\')) && !strtok(lastslash
+1," \n\t")){
4684 *(fullline
+(cont
-len
)+(lastslash
-line
))=' ';
4688 // Not a continuation line. Parse it
4690 scandevice
= ParseConfigLine(conf_entries
, default_conf
, scan_types
, contlineno
, fullline
);
4692 // did we find a scandevice directive?
4695 // did we find a syntax error
4704 // note -- may be zero if syntax of file OK, but no valid entries!
4708 /* Prints the message "=======> VALID ARGUMENTS ARE: <LIST> <=======\n", where
4709 <LIST> is the list of valid arguments for option opt. */
4710 static void PrintValidArgs(char opt
)
4714 PrintOut(LOG_CRIT
, "=======> VALID ARGUMENTS ARE: ");
4715 if (!(s
= GetValidArgList(opt
)))
4716 PrintOut(LOG_CRIT
, "Error constructing argument list for option %c", opt
);
4718 PrintOut(LOG_CRIT
, "%s", (char *)s
);
4719 PrintOut(LOG_CRIT
, " <=======\n");
4723 // Report error and exit if specified path is not absolute.
4724 static void check_abs_path(char option
, const std::string
& path
)
4726 if (path
.empty() || path
[0] == '/')
4731 PrintOut(LOG_CRIT
, "=======> INVALID ARGUMENT TO -%c: %s <=======\n\n", option
, path
.c_str());
4732 PrintOut(LOG_CRIT
, "Error: relative path names are not allowed\n\n");
4737 // Parses input line, prints usage message and
4738 // version/license/copyright messages
4739 static void ParseOpts(int argc
, char **argv
)
4741 // Init default path names
4743 configfile
= SMARTMONTOOLS_SYSCONFDIR
"/smartd.conf";
4744 warning_script
= SMARTMONTOOLS_SMARTDSCRIPTDIR
"/smartd_warning.sh";
4746 std::string exedir
= get_exe_dir();
4747 static std::string configfile_str
= exedir
+ "/smartd.conf";
4748 configfile
= configfile_str
.c_str();
4749 warning_script
= exedir
+ "/smartd_warning.cmd";
4752 // Please update GetValidArgList() if you edit shortopts
4753 static const char shortopts
[] = "c:l:q:dDni:p:r:s:A:B:w:Vh?"
4754 #ifdef HAVE_LIBCAP_NG
4758 // Please update GetValidArgList() if you edit longopts
4759 struct option longopts
[] = {
4760 { "configfile", required_argument
, 0, 'c' },
4761 { "logfacility", required_argument
, 0, 'l' },
4762 { "quit", required_argument
, 0, 'q' },
4763 { "debug", no_argument
, 0, 'd' },
4764 { "showdirectives", no_argument
, 0, 'D' },
4765 { "interval", required_argument
, 0, 'i' },
4767 { "no-fork", no_argument
, 0, 'n' },
4769 { "service", no_argument
, 0, 'n' },
4771 { "pidfile", required_argument
, 0, 'p' },
4772 { "report", required_argument
, 0, 'r' },
4773 { "savestates", required_argument
, 0, 's' },
4774 { "attributelog", required_argument
, 0, 'A' },
4775 { "drivedb", required_argument
, 0, 'B' },
4776 { "warnexec", required_argument
, 0, 'w' },
4777 { "version", no_argument
, 0, 'V' },
4778 { "license", no_argument
, 0, 'V' },
4779 { "copyright", no_argument
, 0, 'V' },
4780 { "help", no_argument
, 0, 'h' },
4781 { "usage", no_argument
, 0, 'h' },
4782 #ifdef HAVE_LIBCAP_NG
4783 { "capabilities", no_argument
, 0, 'C' },
4789 bool badarg
= false;
4790 bool use_default_db
= true; // set false on '-B FILE'
4792 // Parse input options.
4794 while ((optchar
= getopt_long(argc
, argv
, shortopts
, longopts
, NULL
)) != -1) {
4802 if (!strcmp(optarg
, "nodev"))
4804 else if (!strcmp(optarg
, "nodevstartup"))
4805 quit
= QUIT_NODEVSTARTUP
;
4806 else if (!strcmp(optarg
, "never"))
4808 else if (!strcmp(optarg
, "onecheck")) {
4809 quit
= QUIT_ONECHECK
;
4812 else if (!strcmp(optarg
, "showtests")) {
4813 quit
= QUIT_SHOWTESTS
;
4816 else if (!strcmp(optarg
, "errors"))
4822 // set the log facility level
4823 if (!strcmp(optarg
, "daemon"))
4824 facility
=LOG_DAEMON
;
4825 else if (!strcmp(optarg
, "local0"))
4826 facility
=LOG_LOCAL0
;
4827 else if (!strcmp(optarg
, "local1"))
4828 facility
=LOG_LOCAL1
;
4829 else if (!strcmp(optarg
, "local2"))
4830 facility
=LOG_LOCAL2
;
4831 else if (!strcmp(optarg
, "local3"))
4832 facility
=LOG_LOCAL3
;
4833 else if (!strcmp(optarg
, "local4"))
4834 facility
=LOG_LOCAL4
;
4835 else if (!strcmp(optarg
, "local5"))
4836 facility
=LOG_LOCAL5
;
4837 else if (!strcmp(optarg
, "local6"))
4838 facility
=LOG_LOCAL6
;
4839 else if (!strcmp(optarg
, "local7"))
4840 facility
=LOG_LOCAL7
;
4845 // enable debug mode
4850 #ifndef _WIN32 // On Windows, --service is already handled by daemon_main()
4855 // print summary of all valid directives
4861 // Period (time interval) for checking
4862 // strtol will set errno in the event of overflow, so we'll check it.
4864 lchecktime
= strtol(optarg
, &tailptr
, 10);
4865 if (*tailptr
!= '\0' || lchecktime
< 10 || lchecktime
> INT_MAX
|| errno
) {
4868 PrintOut(LOG_CRIT
, "======> INVALID INTERVAL: %s <=======\n", optarg
);
4869 PrintOut(LOG_CRIT
, "======> INTERVAL MUST BE INTEGER BETWEEN %d AND %d <=======\n", 10, INT_MAX
);
4870 PrintOut(LOG_CRIT
, "\nUse smartd -h to get a usage summary\n\n");
4873 checktime
= (int)lchecktime
;
4876 // report IOCTL transactions
4878 int n1
= -1, n2
= -1, len
= strlen(optarg
);
4879 char s
[9+1]; unsigned i
= 1;
4880 sscanf(optarg
, "%9[a-z]%n,%u%n", s
, &n1
, &i
, &n2
);
4881 if (!((n1
== len
|| n2
== len
) && 1 <= i
&& i
<= 4)) {
4883 } else if (!strcmp(s
,"ioctl")) {
4884 ata_debugmode
= scsi_debugmode
= nvme_debugmode
= i
;
4885 } else if (!strcmp(s
,"ataioctl")) {
4887 } else if (!strcmp(s
,"scsiioctl")) {
4889 } else if (!strcmp(s
,"nvmeioctl")) {
4897 // alternate configuration file
4898 if (strcmp(optarg
,"-"))
4899 configfile
= (configfile_alt
= optarg
).c_str();
4900 else // read from stdin
4901 configfile
=configfile_stdin
;
4904 // output file with PID number
4908 // path prefix of persistent state file
4909 state_path_prefix
= optarg
;
4912 // path prefix of attribute log file
4913 attrlog_path_prefix
= optarg
;
4917 const char * path
= optarg
;
4918 if (*path
== '+' && path
[1])
4921 use_default_db
= false;
4922 unsigned char savedebug
= debugmode
; debugmode
= 1;
4923 if (!read_drive_database(path
))
4925 debugmode
= savedebug
;
4929 warning_script
= optarg
;
4932 // print version and CVS info
4934 PrintOut(LOG_INFO
, "%s", format_version_info("smartd", true /*full*/).c_str());
4937 #ifdef HAVE_LIBCAP_NG
4939 // enable capabilities
4940 enable_capabilities
= true;
4944 // help: print summary of command-line options
4952 // unrecognized option
4955 // Point arg to the argument in which this option was found.
4956 arg
= argv
[optind
-1];
4957 // Check whether the option is a long option that doesn't map to -h.
4958 if (arg
[1] == '-' && optchar
!= 'h') {
4959 // Iff optopt holds a valid option then argument must be missing.
4960 if (optopt
&& (strchr(shortopts
, optopt
) != NULL
)) {
4961 PrintOut(LOG_CRIT
, "=======> ARGUMENT REQUIRED FOR OPTION: %s <=======\n",arg
+2);
4962 PrintValidArgs(optopt
);
4964 PrintOut(LOG_CRIT
, "=======> UNRECOGNIZED OPTION: %s <=======\n\n",arg
+2);
4966 PrintOut(LOG_CRIT
, "\nUse smartd --help to get a usage summary\n\n");
4970 // Iff optopt holds a valid option then argument must be missing.
4971 if (strchr(shortopts
, optopt
) != NULL
){
4972 PrintOut(LOG_CRIT
, "=======> ARGUMENT REQUIRED FOR OPTION: %c <=======\n",optopt
);
4973 PrintValidArgs(optopt
);
4975 PrintOut(LOG_CRIT
, "=======> UNRECOGNIZED OPTION: %c <=======\n\n",optopt
);
4977 PrintOut(LOG_CRIT
, "\nUse smartd -h to get a usage summary\n\n");
4984 // Check to see if option had an unrecognized or incorrect argument.
4988 // It would be nice to print the actual option name given by the user
4989 // here, but we just print the short form. Please fix this if you know
4990 // a clean way to do it.
4991 PrintOut(LOG_CRIT
, "=======> INVALID ARGUMENT TO -%c: %s <======= \n", optchar
, optarg
);
4992 PrintValidArgs(optchar
);
4993 PrintOut(LOG_CRIT
, "\nUse smartd -h to get a usage summary\n\n");
4998 // non-option arguments are not allowed
4999 if (argc
> optind
) {
5002 PrintOut(LOG_CRIT
, "=======> UNRECOGNIZED ARGUMENT: %s <=======\n\n", argv
[optind
]);
5003 PrintOut(LOG_CRIT
, "\nUse smartd -h to get a usage summary\n\n");
5007 // no pidfile in debug mode
5008 if (debugmode
&& !pid_file
.empty()) {
5011 PrintOut(LOG_CRIT
, "=======> INVALID CHOICE OF OPTIONS: -d and -p <======= \n\n");
5012 PrintOut(LOG_CRIT
, "Error: pid file %s not written in debug (-d) mode\n\n", pid_file
.c_str());
5018 // absolute path names are required due to chdir('/') after fork().
5019 check_abs_path('p', pid_file
);
5020 check_abs_path('s', state_path_prefix
);
5021 check_abs_path('A', attrlog_path_prefix
);
5025 // Read or init drive database
5027 unsigned char savedebug
= debugmode
; debugmode
= 1;
5028 if (!init_drive_database(use_default_db
))
5030 debugmode
= savedebug
;
5037 // Function we call if no configuration file was found or if the
5038 // SCANDIRECTIVE Directive was found. It makes entries for device
5039 // names returned by scan_smart_devices() in os_OSNAME.cpp
5040 static int MakeConfigEntries(const dev_config
& base_cfg
,
5041 dev_config_vector
& conf_entries
, smart_device_list
& scanned_devs
,
5042 const smart_devtype_list
& types
)
5044 // make list of devices
5045 smart_device_list devlist
;
5046 if (!smi()->scan_smart_devices(devlist
, types
)) {
5047 PrintOut(LOG_CRIT
, "DEVICESCAN failed: %s\n", smi()->get_errmsg());
5051 // if no devices, return
5052 if (devlist
.size() <= 0)
5055 // add empty device slots for existing config entries
5056 while (scanned_devs
.size() < conf_entries
.size())
5057 scanned_devs
.push_back((smart_device
*)0);
5059 // loop over entries to create
5060 for (unsigned i
= 0; i
< devlist
.size(); i
++) {
5061 // Move device pointer
5062 smart_device
* dev
= devlist
.release(i
);
5063 scanned_devs
.push_back(dev
);
5065 // Copy configuration, update device and type name
5066 conf_entries
.push_back(base_cfg
);
5067 dev_config
& cfg
= conf_entries
.back();
5068 cfg
.name
= dev
->get_info().info_name
;
5069 cfg
.dev_name
= dev
->get_info().dev_name
;
5070 cfg
.dev_type
= dev
->get_info().dev_type
;
5073 return devlist
.size();
5076 // Returns negative value (see ParseConfigFile()) if config file
5077 // had errors, else number of entries which may be zero or positive.
5078 static int ReadOrMakeConfigEntries(dev_config_vector
& conf_entries
, smart_device_list
& scanned_devs
)
5080 // parse configuration file configfile (normally /etc/smartd.conf)
5081 smart_devtype_list scan_types
;
5082 int entries
= ParseConfigFile(conf_entries
, scan_types
);
5085 // There was an error reading the configuration file.
5086 conf_entries
.clear();
5088 PrintOut(LOG_CRIT
, "Configuration file %s has fatal syntax errors.\n", configfile
);
5092 // no error parsing config file.
5094 // we did not find a SCANDIRECTIVE and did find valid entries
5095 PrintOut(LOG_INFO
, "Configuration file %s parsed.\n", configfile
);
5097 else if (!conf_entries
.empty()) {
5098 // we found a SCANDIRECTIVE or there was no configuration file so
5099 // scan. Configuration file's last entry contains all options
5101 dev_config first
= conf_entries
.back();
5102 conf_entries
.pop_back();
5105 PrintOut(LOG_INFO
,"Configuration file %s was parsed, found %s, scanning devices\n", configfile
, SCANDIRECTIVE
);
5107 PrintOut(LOG_INFO
,"No configuration file %s found, scanning devices\n", configfile
);
5109 // make config list of devices to search for
5110 MakeConfigEntries(first
, conf_entries
, scanned_devs
, scan_types
);
5112 // warn user if scan table found no devices
5113 if (conf_entries
.empty())
5114 PrintOut(LOG_CRIT
,"In the system's table of devices NO devices found to scan\n");
5117 PrintOut(LOG_CRIT
, "Configuration file %s parsed but has no entries\n", configfile
);
5119 return conf_entries
.size();
5122 // Return true if TYPE contains a RAID drive number
5123 static bool is_raid_type(const char * type
)
5125 if (str_starts_with(type
, "sat,"))
5128 if (sscanf(type
, "%*[^,],%d", &i
) != 1)
5133 // Return true if DEV is already in DEVICES[0..NUMDEVS) or IGNORED[*]
5134 static bool is_duplicate_device(const smart_device
* dev
,
5135 const smart_device_list
& devices
, unsigned numdevs
,
5136 const dev_config_vector
& ignored
)
5138 const smart_device::device_info
& info1
= dev
->get_info();
5139 bool is_raid1
= is_raid_type(info1
.dev_type
.c_str());
5141 for (unsigned i
= 0; i
< numdevs
; i
++) {
5142 const smart_device::device_info
& info2
= devices
.at(i
)->get_info();
5143 // -d TYPE options must match if RAID drive number is specified
5144 if ( info1
.dev_name
== info2
.dev_name
5145 && ( info1
.dev_type
== info2
.dev_type
5146 || !is_raid1
|| !is_raid_type(info2
.dev_type
.c_str())))
5150 for (unsigned i
= 0; i
< ignored
.size(); i
++) {
5151 const dev_config
& cfg2
= ignored
.at(i
);
5152 if ( info1
.dev_name
== cfg2
.dev_name
5153 && ( info1
.dev_type
== cfg2
.dev_type
5154 || !is_raid1
|| !is_raid_type(cfg2
.dev_type
.c_str())))
5160 // Register one device, return false on error
5161 static bool register_device(dev_config
& cfg
, dev_state
& state
, smart_device_auto_ptr
& dev
,
5162 const dev_config_vector
* prev_cfgs
)
5166 // Get device of appropriate type
5167 dev
= smi()->get_smart_device(cfg
.name
.c_str(), cfg
.dev_type
.c_str());
5169 if (cfg
.dev_type
.empty())
5170 PrintOut(LOG_INFO
, "Device: %s, unable to autodetect device type\n", cfg
.name
.c_str());
5172 PrintOut(LOG_INFO
, "Device: %s, unsupported device type '%s'\n", cfg
.name
.c_str(), cfg
.dev_type
.c_str());
5178 // Use device from device scan
5183 smart_device::device_info oldinfo
= dev
->get_info();
5185 // Open with autodetect support, may return 'better' device
5186 dev
.replace( dev
->autodetect_open() );
5188 // Report if type has changed
5189 if (oldinfo
.dev_type
!= dev
->get_dev_type())
5190 PrintOut(LOG_INFO
, "Device: %s, type changed from '%s' to '%s'\n",
5191 cfg
.name
.c_str(), oldinfo
.dev_type
.c_str(), dev
->get_dev_type());
5193 // Return if autodetect_open() failed
5194 if (!dev
->is_open()) {
5195 if (debugmode
|| !scanning
)
5196 PrintOut(LOG_INFO
, "Device: %s, open() failed: %s\n", dev
->get_info_name(), dev
->get_errmsg());
5200 // Update informal name
5201 cfg
.name
= dev
->get_info().info_name
;
5202 PrintOut(LOG_INFO
, "Device: %s, opened\n", cfg
.name
.c_str());
5205 const char * typemsg
;
5206 // register ATA device
5209 status
= ATADeviceScan(cfg
, state
, dev
->to_ata(), prev_cfgs
);
5211 // or register SCSI device
5212 else if (dev
->is_scsi()){
5214 status
= SCSIDeviceScan(cfg
, state
, dev
->to_scsi(), prev_cfgs
);
5216 // or register NVMe device
5217 else if (dev
->is_nvme()) {
5219 status
= NVMeDeviceScan(cfg
, state
, dev
->to_nvme(), prev_cfgs
);
5222 PrintOut(LOG_INFO
, "Device: %s, neither ATA, SCSI nor NVMe device\n", cfg
.name
.c_str());
5227 if (!scanning
|| debugmode
) {
5229 PrintOut(scanning
? LOG_INFO
: LOG_CRIT
,
5230 "Unable to register %s device %s at line %d of file %s\n",
5231 typemsg
, cfg
.name
.c_str(), cfg
.lineno
, configfile
);
5233 PrintOut(LOG_INFO
, "Unable to register %s device %s\n",
5234 typemsg
, cfg
.name
.c_str());
5243 // This function tries devices from conf_entries. Each one that can be
5244 // registered is moved onto the [ata|scsi]devices lists and removed
5245 // from the conf_entries list.
5246 static void RegisterDevices(const dev_config_vector
& conf_entries
, smart_device_list
& scanned_devs
,
5247 dev_config_vector
& configs
, dev_state_vector
& states
, smart_device_list
& devices
)
5249 // start by clearing lists/memory of ALL existing devices
5255 dev_config_vector ignored_entries
;
5256 unsigned numnoscan
= 0;
5257 for (unsigned i
= 0; i
< conf_entries
.size(); i
++){
5259 dev_config cfg
= conf_entries
[i
];
5262 // Store for is_duplicate_device() check and ignore
5263 PrintOut(LOG_INFO
, "Device: %s%s%s%s, ignored\n", cfg
.name
.c_str(),
5264 (!cfg
.dev_type
.empty() ? " [" : ""),
5265 cfg
.dev_type
.c_str(),
5266 (!cfg
.dev_type
.empty() ? "]" : ""));
5267 ignored_entries
.push_back(cfg
);
5271 smart_device_auto_ptr dev
;
5273 // Device may already be detected during devicescan
5274 bool scanning
= false;
5275 if (i
< scanned_devs
.size()) {
5276 dev
= scanned_devs
.release(i
);
5278 // Check for a preceding non-DEVICESCAN entry for the same device
5279 if ( (numnoscan
|| !ignored_entries
.empty())
5280 && is_duplicate_device(dev
.get(), devices
, numnoscan
, ignored_entries
)) {
5281 PrintOut(LOG_INFO
, "Device: %s, duplicate, ignored\n", dev
->get_info_name());
5289 // If scanning, pass dev_idinfo of previous devices for duplicate check
5291 if (!register_device(cfg
, state
, dev
, (scanning
? &configs
: 0))) {
5292 // if device is explictly listed and we can't register it, then
5293 // exit unless the user has specified that the device is removable
5295 if (!(cfg
.removable
|| quit
== QUIT_NEVER
)) {
5296 PrintOut(LOG_CRIT
, "Unable to register device %s (no Directive -d removable). Exiting.\n", cfg
.name
.c_str());
5299 PrintOut(LOG_INFO
, "Device: %s, not available\n", cfg
.name
.c_str());
5300 // Prevent retry of registration
5301 ignored_entries
.push_back(cfg
);
5306 // move onto the list of devices
5307 configs
.push_back(cfg
);
5308 states
.push_back(state
);
5309 devices
.push_back(dev
);
5311 numnoscan
= devices
.size();
5314 init_disable_standby_check(configs
);
5318 // Main program without exception handling
5319 static int main_worker(int argc
, char **argv
)
5321 // Initialize interface
5322 smart_interface::init();
5326 // is it our first pass through?
5327 bool firstpass
= true;
5329 // next time to wake up
5330 time_t wakeuptime
= 0;
5332 // parse input and print header and usage info if needed
5333 ParseOpts(argc
,argv
);
5335 // Configuration for each device
5336 dev_config_vector configs
;
5338 dev_state_vector states
;
5339 // Devices to monitor
5340 smart_device_list devices
;
5342 bool write_states_always
= true;
5344 #ifdef HAVE_LIBCAP_NG
5345 // Drop capabilities
5346 if (enable_capabilities
) {
5347 capng_clear(CAPNG_SELECT_BOTH
);
5348 capng_updatev(CAPNG_ADD
, (capng_type_t
)(CAPNG_EFFECTIVE
|CAPNG_PERMITTED
),
5349 CAP_SYS_ADMIN
, CAP_MKNOD
, CAP_SYS_RAWIO
, -1);
5350 capng_apply(CAPNG_SELECT_BOTH
);
5354 // the main loop of the code
5357 // are we exiting from a signal?
5358 if (caughtsigEXIT
) {
5359 // are we exiting with SIGTERM?
5360 int isterm
=(caughtsigEXIT
==SIGTERM
);
5361 int isquit
=(caughtsigEXIT
==SIGQUIT
);
5362 int isok
=debugmode
?isterm
|| isquit
:isterm
;
5364 PrintOut(isok
?LOG_INFO
:LOG_CRIT
, "smartd received signal %d: %s\n",
5365 caughtsigEXIT
, strsignal(caughtsigEXIT
));
5370 // Write state files
5371 if (!state_path_prefix
.empty())
5372 write_all_dev_states(configs
, states
);
5377 // Should we (re)read the config file?
5378 if (firstpass
|| caughtsigHUP
){
5380 // Write state files
5381 if (!state_path_prefix
.empty())
5382 write_all_dev_states(configs
, states
);
5386 "Signal HUP - rereading configuration file %s\n":
5387 "\a\nSignal INT - rereading configuration file %s (" SIGQUIT_KEYNAME
" quits)\n\n",
5392 dev_config_vector conf_entries
; // Entries read from smartd.conf
5393 smart_device_list scanned_devs
; // Devices found during scan
5394 // (re)reads config file, makes >=0 entries
5395 int entries
= ReadOrMakeConfigEntries(conf_entries
, scanned_devs
);
5398 // checks devices, then moves onto ata/scsi list or deallocates.
5399 RegisterDevices(conf_entries
, scanned_devs
, configs
, states
, devices
);
5400 if (!(configs
.size() == devices
.size() && configs
.size() == states
.size()))
5401 throw std::logic_error("Invalid result from RegisterDevices");
5403 else if ( quit
== QUIT_NEVER
5404 || ((quit
== QUIT_NODEV
|| quit
== QUIT_NODEVSTARTUP
) && !firstpass
)) {
5405 // user has asked to continue on error in configuration file
5407 PrintOut(LOG_INFO
,"Reusing previous configuration\n");
5410 // exit with configuration file error status
5411 return (entries
==-3 ? EXIT_READCONF
: entries
==-2 ? EXIT_NOCONF
: EXIT_BADCONF
);
5415 // Log number of devices we are monitoring...
5416 if (devices
.size() > 0 || quit
== QUIT_NEVER
|| (quit
== QUIT_NODEVSTARTUP
&& !firstpass
)) {
5417 int numata
= 0, numscsi
= 0;
5418 for (unsigned i
= 0; i
< devices
.size(); i
++) {
5419 const smart_device
* dev
= devices
.at(i
);
5422 else if (dev
->is_scsi())
5425 PrintOut(LOG_INFO
,"Monitoring %d ATA/SATA, %d SCSI/SAS and %d NVMe devices\n",
5426 numata
, numscsi
, (int)devices
.size() - numata
- numscsi
);
5429 PrintOut(LOG_INFO
,"Unable to monitor any SMART enabled devices. Try debug (-d) option. Exiting...\n");
5433 if (quit
== QUIT_SHOWTESTS
) {
5434 // user has asked to print test schedule
5435 PrintTestSchedule(configs
, states
, devices
);
5439 #ifdef HAVE_LIBCAP_NG
5440 if (enable_capabilities
) {
5441 for (unsigned i
= 0; i
< configs
.size(); i
++) {
5442 if (!configs
[i
].emailaddress
.empty() || !configs
[i
].emailcmdline
.empty()) {
5443 PrintOut(LOG_WARNING
, "Mail can't be enabled together with --capabilities. All mail will be suppressed.\n");
5453 // Always write state files after (re)configuration
5454 write_states_always
= true;
5457 // check all devices once,
5458 // self tests are not started in first pass unless '-q onecheck' is specified
5459 CheckDevicesOnce(configs
, states
, devices
, firstpass
, (!firstpass
|| quit
== QUIT_ONECHECK
));
5461 // Write state files
5462 if (!state_path_prefix
.empty())
5463 write_all_dev_states(configs
, states
, write_states_always
);
5464 write_states_always
= false;
5466 // Write attribute logs
5467 if (!attrlog_path_prefix
.empty())
5468 write_all_dev_attrlogs(configs
, states
);
5470 // user has asked us to exit after first check
5471 if (quit
== QUIT_ONECHECK
) {
5472 PrintOut(LOG_INFO
,"Started with '-q onecheck' option. All devices sucessfully checked once.\n"
5473 "smartd is exiting (exit status 0)\n");
5477 // fork into background if needed
5478 if (firstpass
&& !debugmode
) {
5482 // set exit and signal handlers, write PID file, set wake-up time
5484 Initialize(&wakeuptime
);
5488 // sleep until next check time, or a signal arrives
5489 wakeuptime
= dosleep(wakeuptime
, write_states_always
);
5496 int main(int argc
, char **argv
)
5498 // Windows: internal main function started direct or by service control manager
5499 static int smartd_main(int argc
, char **argv
)
5504 // Do the real work ...
5505 status
= main_worker(argc
, argv
);
5508 // EXIT(status) arrives here
5511 catch (const std::bad_alloc
& /*ex*/) {
5512 // Memory allocation failed (also thrown by std::operator new)
5513 PrintOut(LOG_CRIT
, "Smartd: Out of memory\n");
5514 status
= EXIT_NOMEM
;
5516 catch (const std::exception
& ex
) {
5517 // Other fatal errors
5518 PrintOut(LOG_CRIT
, "Smartd: Exception: %s\n", ex
.what());
5519 status
= EXIT_BADCODE
;
5522 // Check for remaining device objects
5523 if (smart_device::get_num_objects() != 0) {
5524 PrintOut(LOG_CRIT
, "Smartd: Internal Error: %d device object(s) left at exit.\n",
5525 smart_device::get_num_objects());
5526 status
= EXIT_BADCODE
;
5529 if (status
== EXIT_BADCODE
)
5530 PrintOut(LOG_CRIT
, "Please inform " PACKAGE_BUGREPORT
", including output of smartd -V.\n");
5533 status
= Goodbye(status
);
5536 daemon_winsvc_exitcode
= status
;
5543 // Main function for Windows
5544 int main(int argc
, char **argv
){
5545 // Options for smartd windows service
5546 static const daemon_winsvc_options svc_opts
= {
5547 "--service", // cmd_opt
5548 "smartd", "SmartD Service", // servicename, displayname
5550 "Controls and monitors storage devices using the Self-Monitoring, "
5551 "Analysis and Reporting Technology System (SMART) built into "
5552 "ATA/SATA and SCSI/SAS hard drives and solid-state drives. "
5553 "www.smartmontools.org"
5555 // daemon_main() handles daemon and service specific commands
5556 // and starts smartd_main() direct, from a new process,
5557 // or via service control manager
5558 return daemon_main("smartd", &svc_opts
, smartd_main
, argc
, argv
);