2 * Home page of code is: http://www.smartmontools.org
4 * Copyright (C) 2002-11 Bruce Allen
5 * Copyright (C) 2008-16 Christian Franke
6 * Copyright (C) 2000 Michael Cornwell <cornwell@acm.org>
7 * Copyright (C) 2008 Oliver Bock <brevilo@users.sourceforge.net>
10 * Copyright (C) 2008-15 Christian Franke <smartmontools-support@lists.sourceforge.net>
11 >>>>>>> 3d8ad6fa4529eb02ae1391a1e937bf57aad3fb74
13 * This program is free software; you can redistribute it and/or modify
14 * it under the terms of the GNU General Public License as published by
15 * the Free Software Foundation; either version 2, or (at your option)
18 * You should have received a copy of the GNU General Public License
19 * (for example COPYING); If not, see <http://www.gnu.org/licenses/>.
21 * This code was originally developed as a Senior Thesis by Michael Cornwell
22 * at the Concurrent Systems Laboratory (now part of the Storage Systems
23 * Research Center), Jack Baskin School of Engineering, University of
24 * California, Santa Cruz. http://ssrc.soe.ucsc.edu/
31 // unconditionally included files
33 #include <sys/types.h>
34 #include <sys/stat.h> // umask
49 #include <algorithm> // std::replace()
51 // conditionally included files
61 #pragma warning(disable:4761) // "conversion supplied"
62 typedef unsigned short mode_t
;
65 #include <io.h> // umask()
66 #include <process.h> // getpid()
70 #include <io.h> // setmode()
77 // locally included files
79 #include "dev_interface.h"
80 #include "knowndrives.h"
85 // This is for solaris, where signal() resets the handler to SIG_DFL
86 // after the first signal is caught.
88 #define SIGNALFN sigset
90 #define SIGNALFN signal
94 // fork()/signal()/initd simulation for native Windows
95 #include "daemon_win32.h" // daemon_main/detach/signal()
97 #define SIGNALFN daemon_signal
98 #define strsignal daemon_strsignal
99 #define sleep daemon_sleep
100 // SIGQUIT does not exist, CONTROL-Break signals SIGBREAK.
101 #define SIGQUIT SIGBREAK
102 #define SIGQUIT_KEYNAME "CONTROL-Break"
104 #define SIGQUIT_KEYNAME "CONTROL-\\"
108 const char * smartd_cpp_cvsid
= "$Id: smartd.cpp 4308 2016-04-24 13:36:10Z chrfranke $"
110 #if defined (__SVR4) && defined (__sun)
111 extern "C" int getdomainname(char *, int); // no declaration in header files!
114 const char * smartd_cpp_cvsid
= "$Id: smartd.cpp 4059 2015-04-18 17:01:31Z chrfranke $"
115 >>>>>>> 3d8ad6fa4529eb02ae1391a1e937bf57aad3fb74
118 using namespace smartmontools
;
121 #define EXIT_BADCMD 1 // command line did not parse
122 #define EXIT_BADCONF 2 // syntax error in config file
123 #define EXIT_STARTUP 3 // problem forking daemon
124 #define EXIT_PID 4 // problem creating pid file
125 #define EXIT_NOCONF 5 // config file does not exist
126 #define EXIT_READCONF 6 // config file exists but cannot be read
128 #define EXIT_NOMEM 8 // out of memory
129 #define EXIT_BADCODE 10 // internal error - should NEVER happen
131 #define EXIT_BADDEV 16 // we can't monitor this device
132 #define EXIT_NODEV 17 // no devices to monitor
134 #define EXIT_SIGNAL 254 // abort on signal
137 // command-line: 1=debug mode, 2=print presets
138 static unsigned char debugmode
= 0;
140 // command-line: how long to sleep between checks
141 #define CHECKTIME 1800
142 static int checktime
=CHECKTIME
;
144 // command-line: name of PID file (empty for no pid file)
145 static std::string pid_file
;
147 // command-line: path prefix of persistent state file, empty if no persistence.
148 static std::string state_path_prefix
149 #ifdef SMARTMONTOOLS_SAVESTATES
150 = SMARTMONTOOLS_SAVESTATES
154 // command-line: path prefix of attribute log file, empty if no logs.
155 static std::string attrlog_path_prefix
156 #ifdef SMARTMONTOOLS_ATTRIBUTELOG
157 = SMARTMONTOOLS_ATTRIBUTELOG
161 // configuration file name
162 static const char * configfile
;
163 // configuration file "name" if read from stdin
164 static const char * const configfile_stdin
= "<stdin>";
165 // path of alternate configuration file
166 static std::string configfile_alt
;
168 // warning script file
169 static std::string warning_script
;
171 // command-line: when should we exit?
174 // command-line; this is the default syslog(3) log facility to use.
175 static int facility
=LOG_DAEMON
;
178 // command-line: fork into background?
179 static bool do_fork
=true;
182 #ifdef HAVE_LIBCAP_NG
183 // command-line: enable capabilities?
184 static bool enable_capabilities
= false;
187 // TODO: This smartctl only variable is also used in os_win32.cpp
188 unsigned char failuretest_permissive
= 0;
190 // set to one if we catch a USR1 (check devices now)
191 static volatile int caughtsigUSR1
=0;
194 // set to one if we catch a USR2 (toggle debug mode)
195 static volatile int caughtsigUSR2
=0;
198 // set to one if we catch a HUP (reload config file). In debug mode,
199 // set to two, if we catch INT (also reload config file).
200 static volatile int caughtsigHUP
=0;
202 // set to signal value if we catch INT, QUIT, or TERM
203 static volatile int caughtsigEXIT
=0;
205 // This function prints either to stdout or to the syslog as needed.
206 static void PrintOut(int priority
, const char *fmt
, ...)
207 __attribute_format_printf(2, 3);
209 // Attribute monitoring flags.
210 // See monitor_attr_flags below.
212 MONITOR_IGN_FAILUSE
= 0x01,
213 MONITOR_IGNORE
= 0x02,
214 MONITOR_RAW_PRINT
= 0x04,
216 MONITOR_AS_CRIT
= 0x10,
217 MONITOR_RAW_AS_CRIT
= 0x20,
220 // Array of flags for each attribute.
221 class attribute_flags
225 { memset(m_flags
, 0, sizeof(m_flags
)); }
227 bool is_set(int id
, unsigned char flag
) const
228 { return (0 < id
&& id
< (int)sizeof(m_flags
) && (m_flags
[id
] & flag
)); }
230 void set(int id
, unsigned char flags
)
232 if (0 < id
&& id
< (int)sizeof(m_flags
))
233 m_flags
[id
] |= flags
;
237 unsigned char m_flags
[256];
241 /// Configuration data for a device. Read from smartd.conf.
242 /// Supports copy & assignment and is compatible with STL containers.
245 int lineno
; // Line number of entry in file
246 std::string name
; // Device name (with optional extra info)
247 std::string dev_name
; // Device name (plain, for SMARTD_DEVICE variable)
248 std::string dev_type
; // Device type argument from -d directive, empty if none
249 std::string dev_idinfo
; // Device identify info for warning emails
250 std::string state_file
; // Path of the persistent state file, empty if none
251 std::string attrlog_file
; // Path of the persistent attrlog file, empty if none
252 bool ignore
; // Ignore this entry
253 bool smartcheck
; // Check SMART status
254 bool usagefailed
; // Check for failed Usage Attributes
255 bool prefail
; // Track changes in Prefail Attributes
256 bool usage
; // Track changes in Usage Attributes
257 bool selftest
; // Monitor number of selftest errors
258 bool errorlog
; // Monitor number of ATA errors
259 bool xerrorlog
; // Monitor number of ATA errors (Extended Comprehensive error log)
260 bool offlinests
; // Monitor changes in offline data collection status
261 bool offlinests_ns
; // Disable auto standby if in progress
262 bool selfteststs
; // Monitor changes in self-test execution status
263 bool selfteststs_ns
; // Disable auto standby if in progress
264 bool permissive
; // Ignore failed SMART commands
265 char autosave
; // 1=disable, 2=enable Autosave Attributes
266 char autoofflinetest
; // 1=disable, 2=enable Auto Offline Test
267 firmwarebug_defs firmwarebugs
; // -F directives from drivedb or smartd.conf
268 bool ignorepresets
; // Ignore database of -v options
269 bool showpresets
; // Show database entry for this device
270 bool removable
; // Device may disappear (not be present)
271 char powermode
; // skip check, if disk in idle or standby mode
272 bool powerquiet
; // skip powermode 'skipping checks' message
273 int powerskipmax
; // how many times can be check skipped
274 unsigned char tempdiff
; // Track Temperature changes >= this limit
275 unsigned char tempinfo
, tempcrit
; // Track Temperatures >= these limits as LOG_INFO, LOG_CRIT+mail
276 regular_expression test_regex
; // Regex for scheduled testing
278 // Configuration of email warning messages
279 std::string emailcmdline
; // script to execute, empty if no messages
280 std::string emailaddress
; // email address, or empty
281 unsigned char emailfreq
; // Emails once (1) daily (2) diminishing (3)
282 bool emailtest
; // Send test email?
285 int dev_rpm
; // rotation rate, 0 = unknown, 1 = SSD, >1 = HDD
286 int set_aam
; // disable(-1), enable(1..255->0..254) Automatic Acoustic Management
287 int set_apm
; // disable(-1), enable(2..255->1..254) Advanced Power Management
288 int set_lookahead
; // disable(-1), enable(1) read look-ahead
289 int set_standby
; // set(1..255->0..254) standby timer
290 bool set_security_freeze
; // Freeze ATA security
291 int set_wcache
; // disable(-1), enable(1) write cache
293 bool sct_erc_set
; // set SCT ERC to:
294 unsigned short sct_erc_readtime
; // ERC read time (deciseconds)
295 unsigned short sct_erc_writetime
; // ERC write time (deciseconds)
297 unsigned char curr_pending_id
; // ID of current pending sector count, 0 if none
298 unsigned char offl_pending_id
; // ID of offline uncorrectable sector count, 0 if none
299 bool curr_pending_incr
, offl_pending_incr
; // True if current/offline pending values increase
300 bool curr_pending_set
, offl_pending_set
; // True if '-C', '-U' set in smartd.conf
302 attribute_flags monitor_attr_flags
; // MONITOR_* flags for each attribute
304 ata_vendor_attr_defs attribute_defs
; // -v options
309 dev_config::dev_config()
319 offlinests(false), offlinests_ns(false),
320 selfteststs(false), selfteststs_ns(false),
324 ignorepresets(false),
331 tempinfo(0), tempcrit(0),
335 set_aam(0), set_apm(0),
338 set_security_freeze(false),
341 sct_erc_readtime(0), sct_erc_writetime(0),
342 curr_pending_id(0), offl_pending_id(0),
343 curr_pending_incr(false), offl_pending_incr(false),
344 curr_pending_set(false), offl_pending_set(false)
349 // Number of allowed mail message types
350 static const int SMARTD_NMAIL
= 13;
351 // Type for '-M test' mails (state not persistent)
352 static const int MAILTYPE_TEST
= 0;
353 // TODO: Add const or enum for all mail types.
356 int logged
;// number of times an email has been sent
357 time_t firstsent
;// time first email was sent, as defined by time(2)
358 time_t lastsent
; // time last email was sent, as defined by time(2)
361 : logged(0), firstsent(0), lastsent(0) { }
364 /// Persistent state data for a device.
365 struct persistent_dev_state
367 unsigned char tempmin
, tempmax
; // Min/Max Temperatures
369 unsigned char selflogcount
; // total number of self-test errors
370 unsigned short selfloghour
; // lifetime hours of last self-test error
372 time_t scheduled_test_next_check
; // Time of next check for scheduled self-tests
374 uint64_t selective_test_last_start
; // Start LBA of last scheduled selective self-test
375 uint64_t selective_test_last_end
; // End LBA of last scheduled selective self-test
377 mailinfo maillog
[SMARTD_NMAIL
]; // log info on when mail sent
380 int ataerrorcount
; // Total number of ATA errors
382 // Persistent part of ata_smart_values:
383 struct ata_attribute
{
386 unsigned char worst
; // Byte needed for 'raw64' attribute only.
390 ata_attribute() : id(0), val(0), worst(0), raw(0), resvd(0) { }
392 ata_attribute ata_attributes
[NUMBER_ATA_SMART_ATTRIBUTES
];
396 struct scsi_error_counter_t
{
397 struct scsiErrorCounter errCounter
;
399 scsi_error_counter_t() : found(0)
400 { memset(&errCounter
, 0, sizeof(errCounter
)); }
402 scsi_error_counter_t scsi_error_counters
[3];
404 struct scsi_nonmedium_error_t
{
405 struct scsiNonMediumError nme
;
407 scsi_nonmedium_error_t() : found(0)
408 { memset(&nme
, 0, sizeof(nme
)); }
410 scsi_nonmedium_error_t scsi_nonmedium_error
;
413 uint64_t nvme_err_log_entries
;
415 persistent_dev_state();
418 persistent_dev_state::persistent_dev_state()
419 : tempmin(0), tempmax(0),
422 scheduled_test_next_check(0),
423 selective_test_last_start(0),
424 selective_test_last_end(0),
426 nvme_err_log_entries(0)
430 /// Non-persistent state data for a device.
431 struct temp_dev_state
433 bool must_write
; // true if persistent part should be written
435 bool not_cap_offline
; // true == not capable of offline testing
436 bool not_cap_conveyance
;
439 bool not_cap_selective
;
441 unsigned char temperature
; // last recorded Temperature (in Celsius)
442 time_t tempmin_delay
; // time where Min Temperature tracking will start
444 bool powermodefail
; // true if power mode check failed
445 int powerskipcnt
; // Number of checks skipped due to idle or standby mode
446 int lastpowermodeskipped
; // the last power mode that was skipped
449 unsigned char SmartPageSupported
; // has log sense IE page (0x2f)
450 unsigned char TempPageSupported
; // has log sense temperature page (0xd)
451 unsigned char ReadECounterPageSupported
;
452 unsigned char WriteECounterPageSupported
;
453 unsigned char VerifyECounterPageSupported
;
454 unsigned char NonMediumErrorPageSupported
;
455 unsigned char SuppressReport
; // minimize nuisance reports
456 unsigned char modese_len
; // mode sense/select cmd len: 0 (don't
459 uint64_t num_sectors
; // Number of sectors
460 ata_smart_values smartval
; // SMART data
461 ata_smart_thresholds_pvt smartthres
; // SMART thresholds
462 bool offline_started
; // true if offline data collection was started
463 bool selftest_started
; // true if self-test was started
468 temp_dev_state::temp_dev_state()
470 not_cap_offline(false),
471 not_cap_conveyance(false),
472 not_cap_short(false),
474 not_cap_selective(false),
477 powermodefail(false),
479 lastpowermodeskipped(0),
480 SmartPageSupported(false),
481 TempPageSupported(false),
482 ReadECounterPageSupported(false),
483 WriteECounterPageSupported(false),
484 VerifyECounterPageSupported(false),
485 NonMediumErrorPageSupported(false),
486 SuppressReport(false),
489 offline_started(false),
490 selftest_started(false)
492 memset(&smartval
, 0, sizeof(smartval
));
493 memset(&smartthres
, 0, sizeof(smartthres
));
496 /// Runtime state data for a device.
498 : public persistent_dev_state
,
499 public temp_dev_state
501 void update_persistent_state();
502 void update_temp_state();
505 /// Container for configuration info for each device.
506 typedef std::vector
<dev_config
> dev_config_vector
;
508 /// Container for state info for each device.
509 typedef std::vector
<dev_state
> dev_state_vector
;
511 // Copy ATA attributes to persistent state.
512 void dev_state::update_persistent_state()
514 for (int i
= 0; i
< NUMBER_ATA_SMART_ATTRIBUTES
; i
++) {
515 const ata_smart_attribute
& ta
= smartval
.vendor_attributes
[i
];
516 ata_attribute
& pa
= ata_attributes
[i
];
519 pa
.val
= pa
.worst
= 0; pa
.raw
= 0;
527 | ((uint64_t)ta
.raw
[3] << 24)
528 | ((uint64_t)ta
.raw
[4] << 32)
529 | ((uint64_t)ta
.raw
[5] << 40);
530 pa
.resvd
= ta
.reserv
;
534 // Copy ATA from persistent to temp state.
535 void dev_state::update_temp_state()
537 for (int i
= 0; i
< NUMBER_ATA_SMART_ATTRIBUTES
; i
++) {
538 const ata_attribute
& pa
= ata_attributes
[i
];
539 ata_smart_attribute
& ta
= smartval
.vendor_attributes
[i
];
542 ta
.current
= ta
.worst
= 0;
543 memset(ta
.raw
, 0, sizeof(ta
.raw
));
548 ta
.raw
[0] = (unsigned char) pa
.raw
;
549 ta
.raw
[1] = (unsigned char)(pa
.raw
>> 8);
550 ta
.raw
[2] = (unsigned char)(pa
.raw
>> 16);
551 ta
.raw
[3] = (unsigned char)(pa
.raw
>> 24);
552 ta
.raw
[4] = (unsigned char)(pa
.raw
>> 32);
553 ta
.raw
[5] = (unsigned char)(pa
.raw
>> 40);
554 ta
.reserv
= pa
.resvd
;
558 // Parse a line from a state file.
559 static bool parse_dev_state_line(const char * line
, persistent_dev_state
& state
)
561 static const regular_expression
regex(
563 "((temperature-min)" // (1 (2)
564 "|(temperature-max)" // (3)
565 "|(self-test-errors)" // (4)
566 "|(self-test-last-err-hour)" // (5)
567 "|(scheduled-test-next-check)" // (6)
568 "|(selective-test-last-start)" // (7)
569 "|(selective-test-last-end)" // (8)
570 "|(ata-error-count)" // (9)
571 "|(mail\\.([0-9]+)\\." // (10 (11)
572 "((count)" // (12 (13)
573 "|(first-sent-time)" // (14)
574 "|(last-sent-time)" // (15)
577 "|(ata-smart-attribute\\.([0-9]+)\\." // (16 (17)
585 "|(nvme-err-log-entries)" // (24)
587 " *= *([0-9]+)[ \n]*$", // (25)
591 const int nmatch
= 1+25;
592 regmatch_t match
[nmatch
];
593 if (!regex
.execute(line
, nmatch
, match
))
595 if (match
[nmatch
-1].rm_so
< 0)
598 uint64_t val
= strtoull(line
+ match
[nmatch
-1].rm_so
, (char **)0, 10);
601 if (match
[++m
].rm_so
>= 0)
602 state
.tempmin
= (unsigned char)val
;
603 else if (match
[++m
].rm_so
>= 0)
604 state
.tempmax
= (unsigned char)val
;
605 else if (match
[++m
].rm_so
>= 0)
606 state
.selflogcount
= (unsigned char)val
;
607 else if (match
[++m
].rm_so
>= 0)
608 state
.selfloghour
= (unsigned short)val
;
609 else if (match
[++m
].rm_so
>= 0)
610 state
.scheduled_test_next_check
= (time_t)val
;
611 else if (match
[++m
].rm_so
>= 0)
612 state
.selective_test_last_start
= val
;
613 else if (match
[++m
].rm_so
>= 0)
614 state
.selective_test_last_end
= val
;
615 else if (match
[++m
].rm_so
>= 0)
616 state
.ataerrorcount
= (int)val
;
617 else if (match
[m
+=2].rm_so
>= 0) {
618 int i
= atoi(line
+match
[m
].rm_so
);
619 if (!(0 <= i
&& i
< SMARTD_NMAIL
))
621 if (i
== MAILTYPE_TEST
) // Don't suppress test mails
623 if (match
[m
+=2].rm_so
>= 0)
624 state
.maillog
[i
].logged
= (int)val
;
625 else if (match
[++m
].rm_so
>= 0)
626 state
.maillog
[i
].firstsent
= (time_t)val
;
627 else if (match
[++m
].rm_so
>= 0)
628 state
.maillog
[i
].lastsent
= (time_t)val
;
632 else if (match
[m
+=5+1].rm_so
>= 0) {
633 int i
= atoi(line
+match
[m
].rm_so
);
634 if (!(0 <= i
&& i
< NUMBER_ATA_SMART_ATTRIBUTES
))
636 if (match
[m
+=2].rm_so
>= 0)
637 state
.ata_attributes
[i
].id
= (unsigned char)val
;
638 else if (match
[++m
].rm_so
>= 0)
639 state
.ata_attributes
[i
].val
= (unsigned char)val
;
640 else if (match
[++m
].rm_so
>= 0)
641 state
.ata_attributes
[i
].worst
= (unsigned char)val
;
642 else if (match
[++m
].rm_so
>= 0)
643 state
.ata_attributes
[i
].raw
= val
;
644 else if (match
[++m
].rm_so
>= 0)
645 state
.ata_attributes
[i
].resvd
= (unsigned char)val
;
649 else if (match
[m
+7].rm_so
>= 0)
650 state
.nvme_err_log_entries
= val
;
656 // Read a state file.
657 static bool read_dev_state(const char * path
, persistent_dev_state
& state
)
659 stdio_file
f(path
, "r");
662 pout("Cannot read state file \"%s\"\n", path
);
666 setmode(fileno(f
), O_TEXT
); // Allow files with \r\n
669 persistent_dev_state new_state
;
670 int good
= 0, bad
= 0;
672 while (fgets(line
, sizeof(line
), f
)) {
673 const char * s
= line
+ strspn(line
, " \t");
674 if (!*s
|| *s
== '#')
676 if (!parse_dev_state_line(line
, new_state
))
684 pout("%s: format error\n", path
);
687 pout("%s: %d invalid line(s) ignored\n", path
, bad
);
690 // This sets the values missing in the file to 0.
695 static void write_dev_state_line(FILE * f
, const char * name
, uint64_t val
)
698 fprintf(f
, "%s = %" PRIu64
"\n", name
, val
);
701 static void write_dev_state_line(FILE * f
, const char * name1
, int id
, const char * name2
, uint64_t val
)
704 fprintf(f
, "%s.%d.%s = %" PRIu64
"\n", name1
, id
, name2
, val
);
707 // Write a state file
708 static bool write_dev_state(const char * path
, const persistent_dev_state
& state
)
710 // Rename old "file" to "file~"
711 std::string pathbak
= path
; pathbak
+= '~';
712 unlink(pathbak
.c_str());
713 rename(path
, pathbak
.c_str());
715 stdio_file
f(path
, "w");
717 pout("Cannot create state file \"%s\"\n", path
);
721 fprintf(f
, "# smartd state file\n");
722 write_dev_state_line(f
, "temperature-min", state
.tempmin
);
723 write_dev_state_line(f
, "temperature-max", state
.tempmax
);
724 write_dev_state_line(f
, "self-test-errors", state
.selflogcount
);
725 write_dev_state_line(f
, "self-test-last-err-hour", state
.selfloghour
);
726 write_dev_state_line(f
, "scheduled-test-next-check", state
.scheduled_test_next_check
);
727 write_dev_state_line(f
, "selective-test-last-start", state
.selective_test_last_start
);
728 write_dev_state_line(f
, "selective-test-last-end", state
.selective_test_last_end
);
731 for (i
= 0; i
< SMARTD_NMAIL
; i
++) {
732 if (i
== MAILTYPE_TEST
) // Don't suppress test mails
734 const mailinfo
& mi
= state
.maillog
[i
];
737 write_dev_state_line(f
, "mail", i
, "count", mi
.logged
);
738 write_dev_state_line(f
, "mail", i
, "first-sent-time", mi
.firstsent
);
739 write_dev_state_line(f
, "mail", i
, "last-sent-time", mi
.lastsent
);
743 write_dev_state_line(f
, "ata-error-count", state
.ataerrorcount
);
745 for (i
= 0; i
< NUMBER_ATA_SMART_ATTRIBUTES
; i
++) {
746 const persistent_dev_state::ata_attribute
& pa
= state
.ata_attributes
[i
];
749 write_dev_state_line(f
, "ata-smart-attribute", i
, "id", pa
.id
);
750 write_dev_state_line(f
, "ata-smart-attribute", i
, "val", pa
.val
);
751 write_dev_state_line(f
, "ata-smart-attribute", i
, "worst", pa
.worst
);
752 write_dev_state_line(f
, "ata-smart-attribute", i
, "raw", pa
.raw
);
753 write_dev_state_line(f
, "ata-smart-attribute", i
, "resvd", pa
.resvd
);
757 write_dev_state_line(f
, "nvme-err-log-entries", state
.nvme_err_log_entries
);
762 // Write to the attrlog file
763 static bool write_dev_attrlog(const char * path
, const dev_state
& state
)
765 stdio_file
f(path
, "a");
767 pout("Cannot create attribute log file \"%s\"\n", path
);
772 time_t now
= time(0);
773 struct tm
* tms
= gmtime(&now
);
774 fprintf(f
, "%d-%02d-%02d %02d:%02d:%02d;",
775 1900+tms
->tm_year
, 1+tms
->tm_mon
, tms
->tm_mday
,
776 tms
->tm_hour
, tms
->tm_min
, tms
->tm_sec
);
778 for (int i
= 0; i
< NUMBER_ATA_SMART_ATTRIBUTES
; i
++) {
779 const persistent_dev_state::ata_attribute
& pa
= state
.ata_attributes
[i
];
782 fprintf(f
, "\t%d;%d;%" PRIu64
";", pa
.id
, pa
.val
, pa
.raw
);
785 const struct scsiErrorCounter
* ecp
;
786 const char * pageNames
[3] = {"read", "write", "verify"};
787 for (int k
= 0; k
< 3; ++k
) {
788 if ( !state
.scsi_error_counters
[k
].found
) continue;
789 ecp
= &state
.scsi_error_counters
[k
].errCounter
;
790 fprintf(f
, "\t%s-corr-by-ecc-fast;%" PRIu64
";"
791 "\t%s-corr-by-ecc-delayed;%" PRIu64
";"
792 "\t%s-corr-by-retry;%" PRIu64
";"
793 "\t%s-total-err-corrected;%" PRIu64
";"
794 "\t%s-corr-algorithm-invocations;%" PRIu64
";"
795 "\t%s-gb-processed;%.3f;"
796 "\t%s-total-unc-errors;%" PRIu64
";",
797 pageNames
[k
], ecp
->counter
[0],
798 pageNames
[k
], ecp
->counter
[1],
799 pageNames
[k
], ecp
->counter
[2],
800 pageNames
[k
], ecp
->counter
[3],
801 pageNames
[k
], ecp
->counter
[4],
802 pageNames
[k
], (ecp
->counter
[5] / 1000000000.0),
803 pageNames
[k
], ecp
->counter
[6]);
805 if(state
.scsi_nonmedium_error
.found
&& state
.scsi_nonmedium_error
.nme
.gotPC0
) {
806 fprintf(f
, "\tnon-medium-errors;%" PRIu64
";", state
.scsi_nonmedium_error
.nme
.counterPC0
);
808 // write SCSI current temperature if it is monitored
809 if(state
.TempPageSupported
&& state
.temperature
)
810 fprintf(f
, "\ttemperature;%d;", state
.temperature
);
816 // Write all state files. If write_always is false, don't write
817 // unless must_write is set.
818 static void write_all_dev_states(const dev_config_vector
& configs
,
819 dev_state_vector
& states
,
820 bool write_always
= true)
822 for (unsigned i
= 0; i
< states
.size(); i
++) {
823 const dev_config
& cfg
= configs
.at(i
);
824 if (cfg
.state_file
.empty())
826 dev_state
& state
= states
[i
];
827 if (!write_always
&& !state
.must_write
)
829 if (!write_dev_state(cfg
.state_file
.c_str(), state
))
831 state
.must_write
= false;
832 if (write_always
|| debugmode
)
833 PrintOut(LOG_INFO
, "Device: %s, state written to %s\n",
834 cfg
.name
.c_str(), cfg
.state_file
.c_str());
838 // Write to all attrlog files
839 static void write_all_dev_attrlogs(const dev_config_vector
& configs
,
840 dev_state_vector
& states
)
842 for (unsigned i
= 0; i
< states
.size(); i
++) {
843 const dev_config
& cfg
= configs
.at(i
);
844 if (cfg
.attrlog_file
.empty())
846 dev_state
& state
= states
[i
];
847 write_dev_attrlog(cfg
.attrlog_file
.c_str(), state
);
851 // remove the PID file
852 static void RemovePidFile()
854 if (!pid_file
.empty()) {
855 if (unlink(pid_file
.c_str()))
856 PrintOut(LOG_CRIT
,"Can't unlink PID file %s (%s).\n",
857 pid_file
.c_str(), strerror(errno
));
863 extern "C" { // signal handlers require C-linkage
865 // Note if we catch a SIGUSR1
866 static void USR1handler(int sig
)
874 // Note if we catch a SIGUSR2
875 static void USR2handler(int sig
)
883 // Note if we catch a HUP (or INT in debug mode)
884 static void HUPhandler(int sig
)
893 // signal handler for TERM, QUIT, and INT (if not in debug mode)
894 static void sighandler(int sig
)
903 // Cleanup, print Goodbye message and remove pidfile
904 static int Goodbye(int status
)
906 // delete PID file, if one was created
909 // and this should be the final output from smartd before it exits
910 PrintOut(status
?LOG_CRIT
:LOG_INFO
, "smartd is exiting (exit status %d)\n", status
);
915 // a replacement for setenv() which is not available on all platforms.
916 // Note that the string passed to putenv must not be freed or made
917 // invalid, since a pointer to it is kept by putenv(). This means that
918 // it must either be a static buffer or allocated off the heap. The
919 // string can be freed if the environment variable is redefined via
920 // another call to putenv(). There is no portable way to unset a variable
921 // with putenv(). So we manage the buffer in a static object.
922 // Using setenv() if available is not considered because some
923 // implementations may produce memory leaks.
929 : m_buf((char *)0) { }
931 void set(const char * name
, const char * value
);
936 env_buffer(const env_buffer
&);
937 void operator=(const env_buffer
&);
940 void env_buffer::set(const char * name
, const char * value
)
942 int size
= strlen(name
) + 1 + strlen(value
) + 1;
943 char * newbuf
= new char[size
];
944 snprintf(newbuf
, size
, "%s=%s", name
, value
);
947 throw std::runtime_error("putenv() failed");
949 // This assumes that the same NAME is passed on each call
956 static void MailWarning(const dev_config
& cfg
, dev_state
& state
, int which
, const char *fmt
, ...)
957 __attribute_format_printf(4, 5);
959 // If either address or executable path is non-null then send and log
960 // a warning email, or execute executable
961 static void MailWarning(const dev_config
& cfg
, dev_state
& state
, int which
, const char *fmt
, ...)
963 static const char * const whichfail
[] = {
969 "FailedHealthCheck", // 5
970 "FailedReadSmartData", // 6
971 "FailedReadSmartErrorLog", // 7
972 "FailedReadSmartSelfTestLog", // 8
973 "FailedOpenDevice", // 9
974 "CurrentPendingSector", // 10
975 "OfflineUncorrectableSector", // 11
979 // See if user wants us to send mail
980 if (cfg
.emailaddress
.empty() && cfg
.emailcmdline
.empty())
983 std::string address
= cfg
.emailaddress
;
984 const char * executable
= cfg
.emailcmdline
.c_str();
986 // which type of mail are we sending?
987 mailinfo
* mail
=(state
.maillog
)+which
;
990 if (cfg
.emailfreq
<1 || cfg
.emailfreq
>3) {
991 PrintOut(LOG_CRIT
,"internal error in MailWarning(): cfg.mailwarn->emailfreq=%d\n",cfg
.emailfreq
);
994 if (which
<0 || which
>=SMARTD_NMAIL
|| sizeof(whichfail
)!=SMARTD_NMAIL
*sizeof(char *)) {
995 PrintOut(LOG_CRIT
,"Contact " PACKAGE_BUGREPORT
"; internal error in MailWarning(): which=%d, size=%d\n",
996 which
, (int)sizeof(whichfail
));
1000 // Return if a single warning mail has been sent.
1001 if ((cfg
.emailfreq
==1) && mail
->logged
)
1004 // Return if this is an email test and one has already been sent.
1005 if (which
== 0 && mail
->logged
)
1008 // To decide if to send mail, we need to know what time it is.
1009 time_t epoch
= time(0);
1011 // Return if less than one day has gone by
1012 const int day
= 24*3600;
1013 if (cfg
.emailfreq
==2 && mail
->logged
&& epoch
<(mail
->lastsent
+day
))
1016 // Return if less than 2^(logged-1) days have gone by
1017 if (cfg
.emailfreq
==3 && mail
->logged
) {
1018 int days
= 0x01 << (mail
->logged
- 1);
1020 if (epoch
<(mail
->lastsent
+days
))
1024 #ifdef HAVE_LIBCAP_NG
1025 if (enable_capabilities
) {
1026 PrintOut(LOG_ERR
, "Sending a mail was supressed. "
1027 "Mails can't be send when capabilites are enabled\n");
1032 // record the time of this mail message, and the first mail message
1034 mail
->firstsent
=epoch
;
1035 mail
->lastsent
=epoch
;
1037 // print warning string into message
1041 vsnprintf(message
, sizeof(message
), fmt
, ap
);
1044 // replace commas by spaces to separate recipients
1045 std::replace(address
.begin(), address
.end(), ',', ' ');
1047 // Export information in environment variables that will be useful
1049 static env_buffer env
[12];
1050 env
[0].set("SMARTD_MAILER", executable
);
1051 env
[1].set("SMARTD_MESSAGE", message
);
1052 char dates
[DATEANDEPOCHLEN
];
1053 snprintf(dates
, sizeof(dates
), "%d", mail
->logged
);
1054 env
[2].set("SMARTD_PREVCNT", dates
);
1055 dateandtimezoneepoch(dates
, mail
->firstsent
);
1056 env
[3].set("SMARTD_TFIRST", dates
);
1057 snprintf(dates
, DATEANDEPOCHLEN
,"%d", (int)mail
->firstsent
);
1058 env
[4].set("SMARTD_TFIRSTEPOCH", dates
);
1059 env
[5].set("SMARTD_FAILTYPE", whichfail
[which
]);
1060 env
[6].set("SMARTD_ADDRESS", address
.c_str());
1061 env
[7].set("SMARTD_DEVICESTRING", cfg
.name
.c_str());
1063 // Allow 'smartctl ... -d $SMARTD_DEVICETYPE $SMARTD_DEVICE'
1064 env
[8].set("SMARTD_DEVICETYPE",
1065 (!cfg
.dev_type
.empty() ? cfg
.dev_type
.c_str() : "auto"));
1066 env
[9].set("SMARTD_DEVICE", cfg
.dev_name
.c_str());
1068 env
[10].set("SMARTD_DEVICEINFO", cfg
.dev_idinfo
.c_str());
1070 if (which
) switch (cfg
.emailfreq
) {
1071 case 2: dates
[0] = '1'; dates
[1] = 0; break;
1072 case 3: snprintf(dates
, sizeof(dates
), "%d", (0x01)<<mail
->logged
);
1074 env
[11].set("SMARTD_NEXTDAYS", dates
);
1076 // now construct a command to send this as EMAIL
1078 executable
= "<mail>";
1079 const char * newadd
= (!address
.empty()? address
.c_str() : "<nomailer>");
1080 const char * newwarn
= (which
? "Warning via" : "Test of");
1084 snprintf(command
, sizeof(command
), "%s 2>&1", warning_script
.c_str());
1086 // tell SYSLOG what we are about to do...
1087 PrintOut(LOG_INFO
,"%s %s to %s ...\n",
1088 which
?"Sending warning via":"Executing test of", executable
, newadd
);
1090 // issue the command to send mail or to run the user's executable
1093 if (!(pfp
=popen(command
, "r")))
1094 // failed to popen() mail process
1095 PrintOut(LOG_CRIT
,"%s %s to %s: failed (fork or pipe failed, or no memory) %s\n",
1096 newwarn
, executable
, newadd
, errno
?strerror(errno
):"");
1100 char buffer
[EBUFLEN
];
1102 // if unexpected output on stdout/stderr, null terminate, print, and flush
1103 if ((len
=fread(buffer
, 1, EBUFLEN
, pfp
))) {
1105 int newlen
= len
<EBUFLEN
? len
: EBUFLEN
-1;
1106 buffer
[newlen
]='\0';
1107 PrintOut(LOG_CRIT
,"%s %s to %s produced unexpected output (%s%d bytes) to STDOUT/STDERR: \n%s\n",
1108 newwarn
, executable
, newadd
, len
!=newlen
?"here truncated to ":"", newlen
, buffer
);
1110 // flush pipe if needed
1111 while (fread(buffer
, 1, EBUFLEN
, pfp
) && count
<EBUFLEN
)
1114 // tell user that pipe was flushed, or that something is really wrong
1115 if (count
&& count
<EBUFLEN
)
1116 PrintOut(LOG_CRIT
,"%s %s to %s: flushed remaining STDOUT/STDERR\n",
1117 newwarn
, executable
, newadd
);
1119 PrintOut(LOG_CRIT
,"%s %s to %s: more than 1 MB STDOUT/STDERR flushed, breaking pipe\n",
1120 newwarn
, executable
, newadd
);
1123 // if something went wrong with mail process, print warning
1125 if (-1==(status
=pclose(pfp
)))
1126 PrintOut(LOG_CRIT
,"%s %s to %s: pclose(3) failed %s\n", newwarn
, executable
, newadd
,
1127 errno
?strerror(errno
):"");
1129 // mail process apparently succeeded. Check and report exit status
1130 if (WIFEXITED(status
)) {
1131 // exited 'normally' (but perhaps with nonzero status)
1132 int status8
= WEXITSTATUS(status
);
1134 PrintOut(LOG_CRIT
,"%s %s to %s: failed (32-bit/8-bit exit status: %d/%d) perhaps caught signal %d [%s]\n",
1135 newwarn
, executable
, newadd
, status
, status8
, status8
-128, strsignal(status8
-128));
1137 PrintOut(LOG_CRIT
,"%s %s to %s: failed (32-bit/8-bit exit status: %d/%d)\n",
1138 newwarn
, executable
, newadd
, status
, status8
);
1140 PrintOut(LOG_INFO
,"%s %s to %s: successful\n", newwarn
, executable
, newadd
);
1143 if (WIFSIGNALED(status
))
1144 PrintOut(LOG_INFO
,"%s %s to %s: exited because of uncaught signal %d [%s]\n",
1145 newwarn
, executable
, newadd
, WTERMSIG(status
), strsignal(WTERMSIG(status
)));
1147 // this branch is probably not possible. If subprocess is
1148 // stopped then pclose() should not return.
1149 if (WIFSTOPPED(status
))
1150 PrintOut(LOG_CRIT
,"%s %s to %s: process STOPPED because it caught signal %d [%s]\n",
1151 newwarn
, executable
, newadd
, WSTOPSIG(status
), strsignal(WSTOPSIG(status
)));
1159 snprintf(command
, sizeof(command
), "cmd /c \"%s\"", warning_script
.c_str());
1161 char stdoutbuf
[800]; // < buffer in syslog_win32::vsyslog()
1164 PrintOut(LOG_INFO
,"%s %s to %s ...\n",
1165 (which
?"Sending warning via":"Executing test of"), executable
, newadd
);
1166 rc
= daemon_spawn(command
, "", 0, stdoutbuf
, sizeof(stdoutbuf
));
1167 if (rc
>= 0 && stdoutbuf
[0])
1168 PrintOut(LOG_CRIT
,"%s %s to %s produced unexpected output (%d bytes) to STDOUT/STDERR:\n%s\n",
1169 newwarn
, executable
, newadd
, (int)strlen(stdoutbuf
), stdoutbuf
);
1171 PrintOut(LOG_CRIT
,"%s %s to %s: failed, exit status %d\n",
1172 newwarn
, executable
, newadd
, rc
);
1174 PrintOut(LOG_INFO
,"%s %s to %s: successful\n", newwarn
, executable
, newadd
);
1179 // increment mail sent counter
1183 static void reset_warning_mail(const dev_config
& cfg
, dev_state
& state
, int which
, const char *fmt
, ...)
1184 __attribute_format_printf(4, 5);
1186 static void reset_warning_mail(const dev_config
& cfg
, dev_state
& state
, int which
, const char *fmt
, ...)
1188 if (!(0 <= which
&& which
< SMARTD_NMAIL
))
1191 // Return if no mail sent yet
1192 mailinfo
& mi
= state
.maillog
[which
];
1196 // Format & print message
1200 vsnprintf(msg
, sizeof(msg
), fmt
, ap
);
1203 PrintOut(LOG_INFO
, "Device: %s, %s, warning condition reset after %d email%s\n", cfg
.name
.c_str(),
1204 msg
, mi
.logged
, (mi
.logged
==1 ? "" : "s"));
1206 // Clear mail counter and timestamps
1208 state
.must_write
= true;
1213 // Output multiple lines via separate syslog(3) calls.
1214 static void vsyslog_lines(int priority
, const char * fmt
, va_list ap
)
1216 char buf
[512+EBUFLEN
]; // enough space for exec cmd output in MailWarning()
1217 vsnprintf(buf
, sizeof(buf
), fmt
, ap
);
1219 for (char * p
= buf
, * q
; p
&& *p
; p
= q
) {
1220 if ((q
= strchr(p
, '\n')))
1223 syslog(priority
, "%s\n", p
);
1228 // os_win32/syslog_win32.cpp supports multiple lines.
1229 #define vsyslog_lines vsyslog
1232 // Printing function for watching ataprint commands, or losing them
1233 // [From GLIBC Manual: Since the prototype doesn't specify types for
1234 // optional arguments, in a call to a variadic function the default
1235 // argument promotions are performed on the optional argument
1236 // values. This means the objects of type char or short int (whether
1237 // signed or not) are promoted to either int or unsigned int, as
1239 void pout(const char *fmt
, ...){
1242 // get the correct time in syslog()
1243 FixGlibcTimeZoneBug();
1244 // initialize variable argument list
1246 // in debugmode==1 mode we will print the output from the ataprint.o functions!
1247 if (debugmode
&& debugmode
!= 2) {
1250 if (facility
== LOG_LOCAL1
) // logging to stdout
1253 vfprintf(f
, fmt
, ap
);
1256 // in debugmode==2 mode we print output from knowndrives.o functions
1257 else if (debugmode
==2 || ata_debugmode
|| scsi_debugmode
) {
1258 openlog("smartd", LOG_PID
, facility
);
1259 vsyslog_lines(LOG_INFO
, fmt
, ap
);
1266 // This function prints either to stdout or to the syslog as needed.
1267 static void PrintOut(int priority
, const char *fmt
, ...){
1270 // get the correct time in syslog()
1271 FixGlibcTimeZoneBug();
1272 // initialize variable argument list
1277 if (facility
== LOG_LOCAL1
) // logging to stdout
1280 vfprintf(f
, fmt
, ap
);
1284 openlog("smartd", LOG_PID
, facility
);
1285 vsyslog_lines(priority
, fmt
, ap
);
1292 // Used to warn users about invalid checksums. Called from atacmds.cpp.
1293 void checksumwarning(const char * string
)
1295 pout("Warning! %s error: invalid SMART checksum.\n", string
);
1300 // Wait for the pid file to show up, this makes sure a calling program knows
1301 // that the daemon is really up and running and has a pid to kill it
1302 static bool WaitForPidFile()
1304 int waited
, max_wait
= 10;
1305 struct stat stat_buf
;
1307 if (pid_file
.empty() || debugmode
)
1310 for(waited
= 0; waited
< max_wait
; ++waited
) {
1311 if (!stat(pid_file
.c_str(), &stat_buf
)) {
1321 // Forks new process, closes ALL file descriptors, redirects stdin,
1322 // stdout, and stderr. Not quite daemon(). See
1323 // http://www.linuxjournal.com/article/2335
1324 // for a good description of why we do things this way.
1325 static void DaemonInit()
1331 // flush all buffered streams. Else we might get two copies of open
1332 // streams since both parent and child get copies of the buffers.
1336 if ((pid
=fork()) < 0) {
1338 PrintOut(LOG_CRIT
,"smartd unable to fork daemon process!\n");
1342 // we are the parent process, wait for pid file, then exit cleanly
1343 if(!WaitForPidFile()) {
1344 PrintOut(LOG_CRIT
,"PID file %s didn't show up!\n", pid_file
.c_str());
1350 // from here on, we are the child process.
1353 // Fork one more time to avoid any possibility of having terminals
1354 if ((pid
=fork()) < 0) {
1356 PrintOut(LOG_CRIT
,"smartd unable to fork daemon process!\n");
1360 // we are the parent process -- exit cleanly
1363 // Now we are the child's child...
1366 // close any open file descriptors
1367 for (i
=getdtablesize();i
>=0;--i
)
1370 #define NO_warn_unused_result(cmd) { if (cmd) {} ; }
1372 // redirect any IO attempts to /dev/null for stdin
1373 i
=open("/dev/null",O_RDWR
);
1376 NO_warn_unused_result(dup(i
));
1378 NO_warn_unused_result(dup(i
));
1381 NO_warn_unused_result(chdir("/"));
1384 PrintOut(LOG_INFO
, "smartd has fork()ed into background mode. New PID=%d.\n", (int)getpid());
1388 // No fork() on native Win32
1389 // Detach this process from console
1391 if (daemon_detach("smartd")) {
1392 PrintOut(LOG_CRIT
,"smartd unable to detach from console!\n");
1395 // stdin/out/err now closed if not redirected
1401 // create a PID file containing the current process id
1402 static void WritePidFile()
1404 if (!pid_file
.empty()) {
1405 pid_t pid
= getpid();
1408 old_umask
= umask(0077); // rwx------
1410 // Cygwin: smartd service runs on system account, ensure PID file can be read by admins
1411 old_umask
= umask(0033); // rwxr--r--
1414 stdio_file
f(pid_file
.c_str(), "w");
1416 if (!(f
&& fprintf(f
, "%d\n", (int)pid
) > 0 && f
.close())) {
1417 PrintOut(LOG_CRIT
, "unable to write PID file %s - exiting.\n", pid_file
.c_str());
1420 PrintOut(LOG_INFO
, "file %s written containing PID %d\n", pid_file
.c_str(), (int)pid
);
1424 // Prints header identifying version of code and home
1425 static void PrintHead()
1427 PrintOut(LOG_INFO
, "%s\n", format_version_info("smartd").c_str());
1430 // prints help info for configuration file Directives
1431 static void Directives()
1434 "Configuration file (%s) Directives (after device name):\n"
1435 " -d TYPE Set the device type: auto, ignore, removable,\n"
1437 " -T TYPE Set the tolerance to one of: normal, permissive\n"
1438 " -o VAL Enable/disable automatic offline tests (on/off)\n"
1439 " -S VAL Enable/disable attribute autosave (on/off)\n"
1440 " -n MODE No check if: never, sleep[,N][,q], standby[,N][,q], idle[,N][,q]\n"
1441 " -H Monitor SMART Health Status, report if failed\n"
1442 " -s REG Do Self-Test at time(s) given by regular expression REG\n"
1443 " -l TYPE Monitor SMART log or self-test status:\n"
1444 " error, selftest, xerror, offlinests[,ns], selfteststs[,ns]\n"
1445 " -l scterc,R,W Set SCT Error Recovery Control\n"
1446 " -e Change device setting: aam,[N|off], apm,[N|off], lookahead,[on|off],\n"
1447 " security-freeze, standby,[N|off], wcache,[on|off]\n"
1448 " -f Monitor 'Usage' Attributes, report failures\n"
1449 " -m ADD Send email warning to address ADD\n"
1450 " -M TYPE Modify email warning behavior (see man page)\n"
1451 " -p Report changes in 'Prefailure' Attributes\n"
1452 " -u Report changes in 'Usage' Attributes\n"
1453 " -t Equivalent to -p and -u Directives\n"
1454 " -r ID Also report Raw values of Attribute ID with -p, -u or -t\n"
1455 " -R ID Track changes in Attribute ID Raw value with -p, -u or -t\n"
1456 " -i ID Ignore Attribute ID for -f Directive\n"
1457 " -I ID Ignore Attribute ID for -p, -u or -t Directive\n"
1458 " -C ID[+] Monitor [increases of] Current Pending Sectors in Attribute ID\n"
1459 " -U ID[+] Monitor [increases of] Offline Uncorrectable Sectors in Attribute ID\n"
1460 " -W D,I,C Monitor Temperature D)ifference, I)nformal limit, C)ritical limit\n"
1461 " -v N,ST Modifies labeling of Attribute N (see man page) \n"
1462 " -P TYPE Drive-specific presets: use, ignore, show, showall\n"
1463 " -a Default: -H -f -t -l error -l selftest -l selfteststs -C 197 -U 198\n"
1464 " -F TYPE Use firmware bug workaround:\n"
1466 " # Comment: text after a hash sign is ignored\n"
1467 " \\ Line continuation character\n"
1468 "Attribute ID is a decimal integer 1 <= ID <= 255\n"
1469 "Use ID = 0 to turn off -C and/or -U Directives\n"
1470 "Example: /dev/sda -a\n",
1472 smi()->get_valid_dev_types_str().c_str(),
1473 get_valid_firmwarebug_args());
1476 /* Returns a pointer to a static string containing a formatted list of the valid
1477 arguments to the option opt or NULL on failure. */
1478 static const char *GetValidArgList(char opt
)
1483 return "<PATH_PREFIX>";
1485 return "<FILE_NAME>, -";
1487 return "daemon, local0, local1, local2, local3, local4, local5, local6, local7";
1489 return "nodev, errors, nodevstartup, never, onecheck, showtests";
1491 return "ioctl[,N], ataioctl[,N], scsiioctl[,N], nvmeioctl[,N]";
1495 return "<FILE_NAME>";
1497 return "<INTEGER_SECONDS>";
1503 /* prints help information for command syntax */
1506 PrintOut(LOG_INFO
,"Usage: smartd [options]\n\n");
1507 PrintOut(LOG_INFO
," -A PREFIX, --attributelog=PREFIX\n");
1508 PrintOut(LOG_INFO
," Log ATA attribute information to {PREFIX}MODEL-SERIAL.ata.csv\n");
1509 #ifdef SMARTMONTOOLS_ATTRIBUTELOG
1510 PrintOut(LOG_INFO
," [default is " SMARTMONTOOLS_ATTRIBUTELOG
"MODEL-SERIAL.ata.csv]\n");
1512 PrintOut(LOG_INFO
,"\n");
1513 PrintOut(LOG_INFO
," -B [+]FILE, --drivedb=[+]FILE\n");
1514 PrintOut(LOG_INFO
," Read and replace [add] drive database from FILE\n");
1515 PrintOut(LOG_INFO
," [default is +%s", get_drivedb_path_add());
1516 #ifdef SMARTMONTOOLS_DRIVEDBDIR
1517 PrintOut(LOG_INFO
,"\n");
1518 PrintOut(LOG_INFO
," and then %s", get_drivedb_path_default());
1520 PrintOut(LOG_INFO
,"]\n\n");
1521 PrintOut(LOG_INFO
," -c NAME|-, --configfile=NAME|-\n");
1522 PrintOut(LOG_INFO
," Read configuration file NAME or stdin\n");
1523 PrintOut(LOG_INFO
," [default is %s]\n\n", configfile
);
1524 #ifdef HAVE_LIBCAP_NG
1525 PrintOut(LOG_INFO
," -C, --capabilities\n");
1526 PrintOut(LOG_INFO
," Drop unneeded Linux process capabilities.\n"
1527 " Warning: Mail notification does not work when used.\n\n");
1529 PrintOut(LOG_INFO
," -d, --debug\n");
1530 PrintOut(LOG_INFO
," Start smartd in debug mode\n\n");
1531 PrintOut(LOG_INFO
," -D, --showdirectives\n");
1532 PrintOut(LOG_INFO
," Print the configuration file Directives and exit\n\n");
1533 PrintOut(LOG_INFO
," -h, --help, --usage\n");
1534 PrintOut(LOG_INFO
," Display this help and exit\n\n");
1535 PrintOut(LOG_INFO
," -i N, --interval=N\n");
1536 PrintOut(LOG_INFO
," Set interval between disk checks to N seconds, where N >= 10\n\n");
1537 PrintOut(LOG_INFO
," -l local[0-7], --logfacility=local[0-7]\n");
1539 PrintOut(LOG_INFO
," Use syslog facility local0 - local7 or daemon [default]\n\n");
1541 PrintOut(LOG_INFO
," Log to \"./smartd.log\", stdout, stderr [default is event log]\n\n");
1544 PrintOut(LOG_INFO
," -n, --no-fork\n");
1545 PrintOut(LOG_INFO
," Do not fork into background\n\n");
1547 PrintOut(LOG_INFO
," -p NAME, --pidfile=NAME\n");
1548 PrintOut(LOG_INFO
," Write PID file NAME\n\n");
1549 PrintOut(LOG_INFO
," -q WHEN, --quit=WHEN\n");
1550 PrintOut(LOG_INFO
," Quit on one of: %s\n\n", GetValidArgList('q'));
1551 PrintOut(LOG_INFO
," -r, --report=TYPE\n");
1552 PrintOut(LOG_INFO
," Report transactions for one of: %s\n\n", GetValidArgList('r'));
1553 PrintOut(LOG_INFO
," -s PREFIX, --savestates=PREFIX\n");
1554 PrintOut(LOG_INFO
," Save disk states to {PREFIX}MODEL-SERIAL.TYPE.state\n");
1555 #ifdef SMARTMONTOOLS_SAVESTATES
1556 PrintOut(LOG_INFO
," [default is " SMARTMONTOOLS_SAVESTATES
"MODEL-SERIAL.TYPE.state]\n");
1558 PrintOut(LOG_INFO
,"\n");
1559 PrintOut(LOG_INFO
," -w NAME, --warnexec=NAME\n");
1560 PrintOut(LOG_INFO
," Run executable NAME on warnings\n");
1562 PrintOut(LOG_INFO
," [default is " SMARTMONTOOLS_SMARTDSCRIPTDIR
"/smartd_warning.sh]\n\n");
1564 PrintOut(LOG_INFO
," [default is %s/smartd_warning.cmd]\n\n", get_exe_dir().c_str());
1567 PrintOut(LOG_INFO
," --service\n");
1568 PrintOut(LOG_INFO
," Running as windows service (see man page), install with:\n");
1569 PrintOut(LOG_INFO
," smartd install [options]\n");
1570 PrintOut(LOG_INFO
," Remove service with:\n");
1571 PrintOut(LOG_INFO
," smartd remove\n\n");
1573 PrintOut(LOG_INFO
," -V, --version, --license, --copyright\n");
1574 PrintOut(LOG_INFO
," Print License, Copyright, and version information\n");
1577 static int CloseDevice(smart_device
* device
, const char * name
)
1579 if (!device
->close()){
1580 PrintOut(LOG_INFO
,"Device: %s, %s, close() failed\n", name
, device
->get_errmsg());
1583 // device sucessfully closed
1587 // return true if a char is not allowed in a state file name
1588 static bool not_allowed_in_filename(char c
)
1590 return !( ('0' <= c
&& c
<= '9')
1591 || ('A' <= c
&& c
<= 'Z')
1592 || ('a' <= c
&& c
<= 'z'));
1595 // Read error count from Summary or Extended Comprehensive SMART error log
1596 // Return -1 on error
1597 static int read_ata_error_count(ata_device
* device
, const char * name
,
1598 firmwarebug_defs firmwarebugs
, bool extended
)
1601 ata_smart_errorlog log
;
1602 if (ataReadErrorLog(device
, &log
, firmwarebugs
)){
1603 PrintOut(LOG_INFO
,"Device: %s, Read Summary SMART Error Log failed\n",name
);
1606 return (log
.error_log_pointer
? log
.ata_error_count
: 0);
1609 ata_smart_exterrlog logx
;
1610 if (!ataReadExtErrorLog(device
, &logx
, 0, 1 /*first sector only*/, firmwarebugs
)) {
1611 PrintOut(LOG_INFO
,"Device: %s, Read Extended Comprehensive SMART Error Log failed\n",name
);
1614 // Some disks use the reserved byte as index, see ataprint.cpp.
1615 return (logx
.error_log_index
|| logx
.reserved1
? logx
.device_error_count
: 0);
1619 // returns <0 if problem. Otherwise, bottom 8 bits are the self test
1620 // error count, and top bits are the power-on hours of the last error.
1621 static int SelfTestErrorCount(ata_device
* device
, const char * name
,
1622 firmwarebug_defs firmwarebugs
)
1624 struct ata_smart_selftestlog log
;
1626 if (ataReadSelfTestLog(device
, &log
, firmwarebugs
)){
1627 PrintOut(LOG_INFO
,"Device: %s, Read SMART Self Test Log Failed\n",name
);
1631 // return current number of self-test errors
1632 return ataPrintSmartSelfTestlog(&log
, false, firmwarebugs
);
1635 #define SELFTEST_ERRORCOUNT(x) (x & 0xff)
1636 #define SELFTEST_ERRORHOURS(x) ((x >> 8) & 0xffff)
1638 // Check offline data collection status
1639 static inline bool is_offl_coll_in_progress(unsigned char status
)
1641 return ((status
& 0x7f) == 0x03);
1644 // Check self-test execution status
1645 static inline bool is_self_test_in_progress(unsigned char status
)
1647 return ((status
>> 4) == 0xf);
1650 // Log offline data collection status
1651 static void log_offline_data_coll_status(const char * name
, unsigned char status
)
1654 switch (status
& 0x7f) {
1655 case 0x00: msg
= "was never started"; break;
1656 case 0x02: msg
= "was completed without error"; break;
1657 case 0x03: msg
= "is in progress"; break;
1658 case 0x04: msg
= "was suspended by an interrupting command from host"; break;
1659 case 0x05: msg
= "was aborted by an interrupting command from host"; break;
1660 case 0x06: msg
= "was aborted by the device with a fatal error"; break;
1665 PrintOut(((status
& 0x7f) == 0x06 ? LOG_CRIT
: LOG_INFO
),
1666 "Device: %s, offline data collection %s%s\n", name
, msg
,
1667 ((status
& 0x80) ? " (auto:on)" : ""));
1669 PrintOut(LOG_INFO
, "Device: %s, unknown offline data collection status 0x%02x\n",
1673 // Log self-test execution status
1674 static void log_self_test_exec_status(const char * name
, unsigned char status
)
1677 switch (status
>> 4) {
1678 case 0x0: msg
= "completed without error"; break;
1679 case 0x1: msg
= "was aborted by the host"; break;
1680 case 0x2: msg
= "was interrupted by the host with a reset"; break;
1681 case 0x3: msg
= "could not complete due to a fatal or unknown error"; break;
1682 case 0x4: msg
= "completed with error (unknown test element)"; break;
1683 case 0x5: msg
= "completed with error (electrical test element)"; break;
1684 case 0x6: msg
= "completed with error (servo/seek test element)"; break;
1685 case 0x7: msg
= "completed with error (read test element)"; break;
1686 case 0x8: msg
= "completed with error (handling damage?)"; break;
1691 PrintOut(((status
>> 4) >= 0x4 ? LOG_CRIT
: LOG_INFO
),
1692 "Device: %s, previous self-test %s\n", name
, msg
);
1693 else if ((status
>> 4) == 0xf)
1694 PrintOut(LOG_INFO
, "Device: %s, self-test in progress, %u0%% remaining\n",
1695 name
, status
& 0x0f);
1697 PrintOut(LOG_INFO
, "Device: %s, unknown self-test status 0x%02x\n",
1701 // Check pending sector count id (-C, -U directives).
1702 static bool check_pending_id(const dev_config
& cfg
, const dev_state
& state
,
1703 unsigned char id
, const char * msg
)
1705 // Check attribute index
1706 int i
= ata_find_attr_index(id
, state
.smartval
);
1708 PrintOut(LOG_INFO
, "Device: %s, can't monitor %s count - no Attribute %d\n",
1709 cfg
.name
.c_str(), msg
, id
);
1714 uint64_t rawval
= ata_get_attr_raw_value(state
.smartval
.vendor_attributes
[i
],
1715 cfg
.attribute_defs
);
1716 if (rawval
>= (state
.num_sectors
? state
.num_sectors
: 0xffffffffULL
)) {
1717 PrintOut(LOG_INFO
, "Device: %s, ignoring %s count - bogus Attribute %d value %" PRIu64
" (0x%" PRIx64
")\n",
1718 cfg
.name
.c_str(), msg
, id
, rawval
, rawval
);
1725 // Called by ATA/SCSI/NVMeDeviceScan() after successful device check
1726 static void finish_device_scan(dev_config
& cfg
, dev_state
& state
)
1728 // Set cfg.emailfreq if user hasn't set it
1729 if ((!cfg
.emailaddress
.empty() || !cfg
.emailcmdline
.empty()) && !cfg
.emailfreq
) {
1730 // Avoid that emails are suppressed forever due to state persistence
1731 if (cfg
.state_file
.empty())
1732 cfg
.emailfreq
= 1; // '-M once'
1734 cfg
.emailfreq
= 2; // '-M daily'
1737 // Start self-test regex check now if time was not read from state file
1738 if (!cfg
.test_regex
.empty() && !state
.scheduled_test_next_check
)
1739 state
.scheduled_test_next_check
= time(0);
1742 // Common function to format result message for ATA setting
1743 static void format_set_result_msg(std::string
& msg
, const char * name
, bool ok
,
1744 int set_option
= 0, bool has_value
= false)
1751 else if (set_option
< 0)
1754 msg
+= strprintf(":%d", set_option
-1);
1755 else if (set_option
> 0)
1760 // TODO: Add '-F swapid' directive
1761 const bool fix_swapped_id
= false;
1763 // scan to see what ata devices there are, and if they support SMART
1764 static int ATADeviceScan(dev_config
& cfg
, dev_state
& state
, ata_device
* atadev
)
1767 struct ata_identify_device drive
;
1768 const char *name
= cfg
.name
.c_str();
1771 // Device must be open
1773 // Get drive identity structure
1774 if ((retid
= ata_read_identity(atadev
, &drive
, fix_swapped_id
))) {
1776 // Unable to read Identity structure
1777 PrintOut(LOG_INFO
,"Device: %s, not ATA, no IDENTIFY DEVICE Structure\n",name
);
1779 PrintOut(LOG_INFO
,"Device: %s, packet devices [this device %s] not SMART capable\n",
1780 name
, packetdevicetype(retid
-1));
1781 CloseDevice(atadev
, name
);
1785 // Get drive identity, size and rotation rate (HDD/SSD)
1786 char model
[40+1], serial
[20+1], firmware
[8+1];
1787 ata_format_id_string(model
, drive
.model
, sizeof(model
)-1);
1788 ata_format_id_string(serial
, drive
.serial_no
, sizeof(serial
)-1);
1789 ata_format_id_string(firmware
, drive
.fw_rev
, sizeof(firmware
)-1);
1791 ata_size_info sizes
;
1792 ata_get_size_info(&drive
, sizes
);
1793 state
.num_sectors
= sizes
.sectors
;
1794 cfg
.dev_rpm
= ata_get_rotation_rate(&drive
);
1796 char wwn
[30]; wwn
[0] = 0;
1797 unsigned oui
= 0; uint64_t unique_id
= 0;
1798 int naa
= ata_get_wwn(&drive
, oui
, unique_id
);
1800 snprintf(wwn
, sizeof(wwn
), "WWN:%x-%06x-%09" PRIx64
", ", naa
, oui
, unique_id
);
1802 // Format device id string for warning emails
1804 cfg
.dev_idinfo
= strprintf("%s, S/N:%s, %sFW:%s, %s", model
, serial
, wwn
, firmware
,
1805 format_capacity(cap
, sizeof(cap
), sizes
.capacity
, "."));
1807 PrintOut(LOG_INFO
, "Device: %s, %s\n", name
, cfg
.dev_idinfo
.c_str());
1809 // Show if device in database, and use preset vendor attribute
1810 // options unless user has requested otherwise.
1811 if (cfg
.ignorepresets
)
1812 PrintOut(LOG_INFO
, "Device: %s, smartd database not searched (Directive: -P ignore).\n", name
);
1814 // Apply vendor specific presets, print warning if present
1815 const drive_settings
* dbentry
= lookup_drive_apply_presets(
1816 &drive
, cfg
.attribute_defs
, cfg
.firmwarebugs
);
1818 PrintOut(LOG_INFO
, "Device: %s, not found in smartd database.\n", name
);
1820 PrintOut(LOG_INFO
, "Device: %s, found in smartd database%s%s\n",
1821 name
, (*dbentry
->modelfamily
? ": " : "."), (*dbentry
->modelfamily
? dbentry
->modelfamily
: ""));
1822 if (*dbentry
->warningmsg
)
1823 PrintOut(LOG_CRIT
, "Device: %s, WARNING: %s\n", name
, dbentry
->warningmsg
);
1827 // Check for ATA Security LOCK
1828 unsigned short word128
= drive
.words088_255
[128-88];
1829 bool locked
= ((word128
& 0x0007) == 0x0007); // LOCKED|ENABLED|SUPPORTED
1831 PrintOut(LOG_INFO
, "Device: %s, ATA Security is **LOCKED**\n", name
);
1833 // Set default '-C 197[+]' if no '-C ID' is specified.
1834 if (!cfg
.curr_pending_set
)
1835 cfg
.curr_pending_id
= get_unc_attr_id(false, cfg
.attribute_defs
, cfg
.curr_pending_incr
);
1836 // Set default '-U 198[+]' if no '-U ID' is specified.
1837 if (!cfg
.offl_pending_set
)
1838 cfg
.offl_pending_id
= get_unc_attr_id(true, cfg
.attribute_defs
, cfg
.offl_pending_incr
);
1840 // If requested, show which presets would be used for this drive
1841 if (cfg
.showpresets
) {
1842 int savedebugmode
=debugmode
;
1843 PrintOut(LOG_INFO
, "Device %s: presets are:\n", name
);
1846 show_presets(&drive
);
1847 debugmode
=savedebugmode
;
1850 // see if drive supports SMART
1851 supported
=ataSmartSupport(&drive
);
1854 // drive does NOT support SMART
1855 PrintOut(LOG_INFO
,"Device: %s, lacks SMART capability\n",name
);
1857 // can't tell if drive supports SMART
1858 PrintOut(LOG_INFO
,"Device: %s, ATA IDENTIFY DEVICE words 82-83 don't specify if SMART capable.\n",name
);
1860 // should we proceed anyway?
1861 if (cfg
.permissive
) {
1862 PrintOut(LOG_INFO
,"Device: %s, proceeding since '-T permissive' Directive given.\n",name
);
1865 PrintOut(LOG_INFO
,"Device: %s, to proceed anyway, use '-T permissive' Directive.\n",name
);
1866 CloseDevice(atadev
, name
);
1871 if (ataEnableSmart(atadev
)) {
1872 // Enable SMART command has failed
1873 PrintOut(LOG_INFO
,"Device: %s, could not enable SMART capability\n",name
);
1875 if (ataIsSmartEnabled(&drive
) <= 0) {
1876 CloseDevice(atadev
, name
);
1879 PrintOut(LOG_INFO
, "Device: %s, proceeding since SMART is already enabled\n", name
);
1882 // disable device attribute autosave...
1883 if (cfg
.autosave
==1) {
1884 if (ataDisableAutoSave(atadev
))
1885 PrintOut(LOG_INFO
,"Device: %s, could not disable SMART Attribute Autosave.\n",name
);
1887 PrintOut(LOG_INFO
,"Device: %s, disabled SMART Attribute Autosave.\n",name
);
1890 // or enable device attribute autosave
1891 if (cfg
.autosave
==2) {
1892 if (ataEnableAutoSave(atadev
))
1893 PrintOut(LOG_INFO
,"Device: %s, could not enable SMART Attribute Autosave.\n",name
);
1895 PrintOut(LOG_INFO
,"Device: %s, enabled SMART Attribute Autosave.\n",name
);
1898 // capability check: SMART status
1899 if (cfg
.smartcheck
&& ataSmartStatus2(atadev
) == -1) {
1900 PrintOut(LOG_INFO
,"Device: %s, not capable of SMART Health Status check\n",name
);
1901 cfg
.smartcheck
= false;
1904 // capability check: Read smart values and thresholds. Note that
1905 // smart values are ALSO needed even if we ONLY want to know if the
1906 // device is self-test log or error-log capable! After ATA-5, this
1907 // information was ALSO reproduced in the IDENTIFY DEVICE response,
1908 // but sadly not for ATA-5. Sigh.
1910 // do we need to get SMART data?
1911 bool smart_val_ok
= false;
1912 if ( cfg
.autoofflinetest
|| cfg
.selftest
1913 || cfg
.errorlog
|| cfg
.xerrorlog
1914 || cfg
.offlinests
|| cfg
.selfteststs
1915 || cfg
.usagefailed
|| cfg
.prefail
|| cfg
.usage
1916 || cfg
.tempdiff
|| cfg
.tempinfo
|| cfg
.tempcrit
1917 || cfg
.curr_pending_id
|| cfg
.offl_pending_id
) {
1919 if (ataReadSmartValues(atadev
, &state
.smartval
)) {
1920 PrintOut(LOG_INFO
, "Device: %s, Read SMART Values failed\n", name
);
1921 cfg
.usagefailed
= cfg
.prefail
= cfg
.usage
= false;
1922 cfg
.tempdiff
= cfg
.tempinfo
= cfg
.tempcrit
= 0;
1923 cfg
.curr_pending_id
= cfg
.offl_pending_id
= 0;
1926 smart_val_ok
= true;
1927 if (ataReadSmartThresholds(atadev
, &state
.smartthres
)) {
1928 PrintOut(LOG_INFO
, "Device: %s, Read SMART Thresholds failed%s\n",
1929 name
, (cfg
.usagefailed
? ", ignoring -f Directive" : ""));
1930 cfg
.usagefailed
= false;
1931 // Let ata_get_attr_state() return ATTRSTATE_NO_THRESHOLD:
1932 memset(&state
.smartthres
, 0, sizeof(state
.smartthres
));
1936 // see if the necessary Attribute is there to monitor offline or
1937 // current pending sectors or temperature
1938 if ( cfg
.curr_pending_id
1939 && !check_pending_id(cfg
, state
, cfg
.curr_pending_id
,
1940 "Current_Pending_Sector"))
1941 cfg
.curr_pending_id
= 0;
1943 if ( cfg
.offl_pending_id
1944 && !check_pending_id(cfg
, state
, cfg
.offl_pending_id
,
1945 "Offline_Uncorrectable"))
1946 cfg
.offl_pending_id
= 0;
1948 if ( (cfg
.tempdiff
|| cfg
.tempinfo
|| cfg
.tempcrit
)
1949 && !ata_return_temperature_value(&state
.smartval
, cfg
.attribute_defs
)) {
1950 PrintOut(LOG_INFO
, "Device: %s, can't monitor Temperature, ignoring -W %d,%d,%d\n",
1951 name
, cfg
.tempdiff
, cfg
.tempinfo
, cfg
.tempcrit
);
1952 cfg
.tempdiff
= cfg
.tempinfo
= cfg
.tempcrit
= 0;
1955 // Report ignored '-r' or '-R' directives
1956 for (int id
= 1; id
<= 255; id
++) {
1957 if (cfg
.monitor_attr_flags
.is_set(id
, MONITOR_RAW_PRINT
)) {
1958 char opt
= (!cfg
.monitor_attr_flags
.is_set(id
, MONITOR_RAW
) ? 'r' : 'R');
1959 const char * excl
= (cfg
.monitor_attr_flags
.is_set(id
,
1960 (opt
== 'r' ? MONITOR_AS_CRIT
: MONITOR_RAW_AS_CRIT
)) ? "!" : "");
1962 int idx
= ata_find_attr_index(id
, state
.smartval
);
1964 PrintOut(LOG_INFO
,"Device: %s, no Attribute %d, ignoring -%c %d%s\n", name
, id
, opt
, id
, excl
);
1966 bool prefail
= !!ATTRIBUTE_FLAGS_PREFAILURE(state
.smartval
.vendor_attributes
[idx
].flags
);
1967 if (!((prefail
&& cfg
.prefail
) || (!prefail
&& cfg
.usage
)))
1968 PrintOut(LOG_INFO
,"Device: %s, not monitoring %s Attributes, ignoring -%c %d%s\n", name
,
1969 (prefail
? "Prefailure" : "Usage"), opt
, id
, excl
);
1975 // enable/disable automatic on-line testing
1976 if (cfg
.autoofflinetest
) {
1977 // is this an enable or disable request?
1978 const char *what
=(cfg
.autoofflinetest
==1)?"disable":"enable";
1980 PrintOut(LOG_INFO
,"Device: %s, could not %s SMART Automatic Offline Testing.\n",name
, what
);
1982 // if command appears unsupported, issue a warning...
1983 if (!isSupportAutomaticTimer(&state
.smartval
))
1984 PrintOut(LOG_INFO
,"Device: %s, SMART Automatic Offline Testing unsupported...\n",name
);
1985 // ... but then try anyway
1986 if ((cfg
.autoofflinetest
==1)?ataDisableAutoOffline(atadev
):ataEnableAutoOffline(atadev
))
1987 PrintOut(LOG_INFO
,"Device: %s, %s SMART Automatic Offline Testing failed.\n", name
, what
);
1989 PrintOut(LOG_INFO
,"Device: %s, %sd SMART Automatic Offline Testing.\n", name
, what
);
1993 // Read log directories if required for capability check
1994 ata_smart_log_directory smart_logdir
, gp_logdir
;
1995 bool smart_logdir_ok
= false, gp_logdir_ok
= false;
1997 if ( isGeneralPurposeLoggingCapable(&drive
)
1998 && (cfg
.errorlog
|| cfg
.selftest
)
1999 && !cfg
.firmwarebugs
.is_set(BUG_NOLOGDIR
)) {
2000 if (!ataReadLogDirectory(atadev
, &smart_logdir
, false))
2001 smart_logdir_ok
= true;
2004 if (cfg
.xerrorlog
&& !cfg
.firmwarebugs
.is_set(BUG_NOLOGDIR
)) {
2005 if (!ataReadLogDirectory(atadev
, &gp_logdir
, true))
2006 gp_logdir_ok
= true;
2009 // capability check: self-test-log
2010 state
.selflogcount
= 0; state
.selfloghour
= 0;
2013 if (!( cfg
.permissive
2014 || ( smart_logdir_ok
&& smart_logdir
.entry
[0x06-1].numsectors
)
2015 || (!smart_logdir_ok
&& smart_val_ok
&& isSmartTestLogCapable(&state
.smartval
, &drive
)))) {
2016 PrintOut(LOG_INFO
, "Device: %s, no SMART Self-test Log, ignoring -l selftest (override with -T permissive)\n", name
);
2017 cfg
.selftest
= false;
2019 else if ((retval
= SelfTestErrorCount(atadev
, name
, cfg
.firmwarebugs
)) < 0) {
2020 PrintOut(LOG_INFO
, "Device: %s, no SMART Self-test Log, ignoring -l selftest\n", name
);
2021 cfg
.selftest
= false;
2024 state
.selflogcount
=SELFTEST_ERRORCOUNT(retval
);
2025 state
.selfloghour
=SELFTEST_ERRORHOURS(retval
);
2029 // capability check: ATA error log
2030 state
.ataerrorcount
= 0;
2033 if (!( cfg
.permissive
2034 || ( smart_logdir_ok
&& smart_logdir
.entry
[0x01-1].numsectors
)
2035 || (!smart_logdir_ok
&& smart_val_ok
&& isSmartErrorLogCapable(&state
.smartval
, &drive
)))) {
2036 PrintOut(LOG_INFO
, "Device: %s, no SMART Error Log, ignoring -l error (override with -T permissive)\n", name
);
2037 cfg
.errorlog
= false;
2039 else if ((errcnt1
= read_ata_error_count(atadev
, name
, cfg
.firmwarebugs
, false)) < 0) {
2040 PrintOut(LOG_INFO
, "Device: %s, no SMART Error Log, ignoring -l error\n", name
);
2041 cfg
.errorlog
= false;
2044 state
.ataerrorcount
= errcnt1
;
2047 if (cfg
.xerrorlog
) {
2049 if (!( cfg
.permissive
|| cfg
.firmwarebugs
.is_set(BUG_NOLOGDIR
)
2050 || (gp_logdir_ok
&& gp_logdir
.entry
[0x03-1].numsectors
) )) {
2051 PrintOut(LOG_INFO
, "Device: %s, no Extended Comprehensive SMART Error Log, ignoring -l xerror (override with -T permissive)\n",
2053 cfg
.xerrorlog
= false;
2055 else if ((errcnt2
= read_ata_error_count(atadev
, name
, cfg
.firmwarebugs
, true)) < 0) {
2056 PrintOut(LOG_INFO
, "Device: %s, no Extended Comprehensive SMART Error Log, ignoring -l xerror\n", name
);
2057 cfg
.xerrorlog
= false;
2059 else if (cfg
.errorlog
&& state
.ataerrorcount
!= errcnt2
) {
2060 PrintOut(LOG_INFO
, "Device: %s, SMART Error Logs report different error counts: %d != %d\n",
2061 name
, state
.ataerrorcount
, errcnt2
);
2062 // Record max error count
2063 if (errcnt2
> state
.ataerrorcount
)
2064 state
.ataerrorcount
= errcnt2
;
2067 state
.ataerrorcount
= errcnt2
;
2070 // capability check: self-test and offline data collection status
2071 if (cfg
.offlinests
|| cfg
.selfteststs
) {
2072 if (!(cfg
.permissive
|| (smart_val_ok
&& state
.smartval
.offline_data_collection_capability
))) {
2074 PrintOut(LOG_INFO
, "Device: %s, no SMART Offline Data Collection capability, ignoring -l offlinests (override with -T permissive)\n", name
);
2075 if (cfg
.selfteststs
)
2076 PrintOut(LOG_INFO
, "Device: %s, no SMART Self-test capability, ignoring -l selfteststs (override with -T permissive)\n", name
);
2077 cfg
.offlinests
= cfg
.selfteststs
= false;
2081 // capabilities check -- does it support powermode?
2082 if (cfg
.powermode
) {
2083 int powermode
= ataCheckPowerMode(atadev
);
2085 if (-1 == powermode
) {
2086 PrintOut(LOG_CRIT
, "Device: %s, no ATA CHECK POWER STATUS support, ignoring -n Directive\n", name
);
2089 else if (powermode
!=0x00 && powermode
!=0x01
2090 && powermode
!=0x40 && powermode
!=0x41
2091 && powermode
!=0x80 && powermode
!=0x81 && powermode
!=0x82 && powermode
!=0x83
2092 && powermode
!=0xff) {
2093 PrintOut(LOG_CRIT
, "Device: %s, CHECK POWER STATUS returned %d, not ATA compliant, ignoring -n Directive\n",
2099 // Apply ATA settings
2103 format_set_result_msg(msg
, "AAM", (cfg
.set_aam
> 0 ?
2104 ata_set_features(atadev
, ATA_ENABLE_AAM
, cfg
.set_aam
-1) :
2105 ata_set_features(atadev
, ATA_DISABLE_AAM
)), cfg
.set_aam
, true);
2108 format_set_result_msg(msg
, "APM", (cfg
.set_apm
> 0 ?
2109 ata_set_features(atadev
, ATA_ENABLE_APM
, cfg
.set_apm
-1) :
2110 ata_set_features(atadev
, ATA_DISABLE_APM
)), cfg
.set_apm
, true);
2112 if (cfg
.set_lookahead
)
2113 format_set_result_msg(msg
, "Rd-ahead", ata_set_features(atadev
,
2114 (cfg
.set_lookahead
> 0 ? ATA_ENABLE_READ_LOOK_AHEAD
: ATA_DISABLE_READ_LOOK_AHEAD
)),
2118 format_set_result_msg(msg
, "Wr-cache", ata_set_features(atadev
,
2119 (cfg
.set_wcache
> 0? ATA_ENABLE_WRITE_CACHE
: ATA_DISABLE_WRITE_CACHE
)), cfg
.set_wcache
);
2121 if (cfg
.set_security_freeze
)
2122 format_set_result_msg(msg
, "Security freeze",
2123 ata_nodata_command(atadev
, ATA_SECURITY_FREEZE_LOCK
));
2125 if (cfg
.set_standby
)
2126 format_set_result_msg(msg
, "Standby",
2127 ata_nodata_command(atadev
, ATA_IDLE
, cfg
.set_standby
-1), cfg
.set_standby
, true);
2129 // Report as one log entry
2131 PrintOut(LOG_INFO
, "Device: %s, ATA settings applied: %s\n", name
, msg
.c_str());
2133 // set SCT Error Recovery Control if requested
2134 if (cfg
.sct_erc_set
) {
2135 if (!isSCTErrorRecoveryControlCapable(&drive
))
2136 PrintOut(LOG_INFO
, "Device: %s, no SCT Error Recovery Control support, ignoring -l scterc\n",
2139 PrintOut(LOG_INFO
, "Device: %s, no SCT support if ATA Security is LOCKED, ignoring -l scterc\n",
2141 else if ( ataSetSCTErrorRecoveryControltime(atadev
, 1, cfg
.sct_erc_readtime
)
2142 || ataSetSCTErrorRecoveryControltime(atadev
, 2, cfg
.sct_erc_writetime
))
2143 PrintOut(LOG_INFO
, "Device: %s, set of SCT Error Recovery Control failed\n", name
);
2145 PrintOut(LOG_INFO
, "Device: %s, SCT Error Recovery Control set to: Read: %u, Write: %u\n",
2146 name
, cfg
.sct_erc_readtime
, cfg
.sct_erc_writetime
);
2149 // If no tests available or selected, return
2150 if (!( cfg
.smartcheck
|| cfg
.selftest
2151 || cfg
.errorlog
|| cfg
.xerrorlog
2152 || cfg
.offlinests
|| cfg
.selfteststs
2153 || cfg
.usagefailed
|| cfg
.prefail
|| cfg
.usage
2154 || cfg
.tempdiff
|| cfg
.tempinfo
|| cfg
.tempcrit
)) {
2155 CloseDevice(atadev
, name
);
2159 // tell user we are registering device
2160 PrintOut(LOG_INFO
,"Device: %s, is SMART capable. Adding to \"monitor\" list.\n",name
);
2162 // close file descriptor
2163 CloseDevice(atadev
, name
);
2165 if (!state_path_prefix
.empty() || !attrlog_path_prefix
.empty()) {
2166 // Build file name for state file
2167 std::replace_if(model
, model
+strlen(model
), not_allowed_in_filename
, '_');
2168 std::replace_if(serial
, serial
+strlen(serial
), not_allowed_in_filename
, '_');
2169 if (!state_path_prefix
.empty()) {
2170 cfg
.state_file
= strprintf("%s%s-%s.ata.state", state_path_prefix
.c_str(), model
, serial
);
2171 // Read previous state
2172 if (read_dev_state(cfg
.state_file
.c_str(), state
)) {
2173 PrintOut(LOG_INFO
, "Device: %s, state read from %s\n", name
, cfg
.state_file
.c_str());
2174 // Copy ATA attribute values to temp state
2175 state
.update_temp_state();
2178 if (!attrlog_path_prefix
.empty())
2179 cfg
.attrlog_file
= strprintf("%s%s-%s.ata.csv", attrlog_path_prefix
.c_str(), model
, serial
);
2182 finish_device_scan(cfg
, state
);
2187 // on success, return 0. On failure, return >0. Never return <0,
2189 static int SCSIDeviceScan(dev_config
& cfg
, dev_state
& state
, scsi_device
* scsidev
)
2191 int err
, req_len
, avail_len
, version
, len
;
2192 const char *device
= cfg
.name
.c_str();
2193 struct scsi_iec_mode_page iec
;
2197 char lu_id
[64], serial
[256], vendor
[40], model
[40];
2199 // Device must be open
2200 memset(inqBuf
, 0, 96);
2202 if ((err
= scsiStdInquiry(scsidev
, inqBuf
, req_len
))) {
2203 /* Marvell controllers fail on a 36 bytes StdInquiry, but 64 suffices */
2205 if ((err
= scsiStdInquiry(scsidev
, inqBuf
, req_len
))) {
2206 PrintOut(LOG_INFO
, "Device: %s, Both 36 and 64 byte INQUIRY failed; "
2207 "skip device\n", device
);
2211 version
= (inqBuf
[2] & 0x7f); /* Accept old ISO/IEC 9316:1995 variants */
2213 avail_len
= inqBuf
[4] + 5;
2214 len
= (avail_len
< req_len
) ? avail_len
: req_len
;
2216 PrintOut(LOG_INFO
, "Device: %s, INQUIRY response less than 36 bytes; "
2217 "skip device\n", device
);
2221 int pdt
= inqBuf
[0] & 0x1f;
2223 if (! ((0 == pdt
) || (4 == pdt
) || (5 == pdt
) || (7 == pdt
) ||
2225 PrintOut(LOG_INFO
, "Device: %s, not a disk like device [PDT=0x%x], "
2226 "skip\n", device
, pdt
);
2230 if (supported_vpd_pages_p
) {
2231 delete supported_vpd_pages_p
;
2232 supported_vpd_pages_p
= NULL
;
2234 supported_vpd_pages_p
= new supported_vpd_pages(scsidev
);
2237 if ((version
>= 0x3) && (version
< 0x8)) {
2239 if (0 == scsiInquiryVpd(scsidev
, SCSI_VPD_DEVICE_IDENTIFICATION
,
2240 vpdBuf
, sizeof(vpdBuf
))) {
2242 scsi_decode_lu_dev_id(vpdBuf
+ 4, len
, lu_id
, sizeof(lu_id
), NULL
);
2246 if (0 == scsiInquiryVpd(scsidev
, SCSI_VPD_UNIT_SERIAL_NUMBER
,
2247 vpdBuf
, sizeof(vpdBuf
))) {
2249 vpdBuf
[4 + len
] = '\0';
2250 scsi_format_id_string(serial
, (const unsigned char *)&vpdBuf
[4], len
);
2253 unsigned int lb_size
;
2255 uint64_t capacity
= scsiGetSize(scsidev
, &lb_size
, NULL
);
2258 format_capacity(si_str
, sizeof(si_str
), capacity
, ".");
2262 // Format device id string for warning emails
2263 cfg
.dev_idinfo
= strprintf("[%.8s %.16s %.4s]%s%s%s%s%s%s",
2264 (char *)&inqBuf
[8], (char *)&inqBuf
[16], (char *)&inqBuf
[32],
2265 (lu_id
[0] ? ", lu id: " : ""), (lu_id
[0] ? lu_id
: ""),
2266 (serial
[0] ? ", S/N: " : ""), (serial
[0] ? serial
: ""),
2267 (si_str
[0] ? ", " : ""), (si_str
[0] ? si_str
: ""));
2269 // format "model" string
2270 scsi_format_id_string(vendor
, (const unsigned char *)&inqBuf
[8], 8);
2271 scsi_format_id_string(model
, (const unsigned char *)&inqBuf
[16], 16);
2272 PrintOut(LOG_INFO
, "Device: %s, %s\n", device
, cfg
.dev_idinfo
.c_str());
2274 // check that device is ready for commands. IE stores its stuff on
2276 if ((err
= scsiTestUnitReady(scsidev
))) {
2277 if (SIMPLE_ERR_NOT_READY
== err
)
2278 PrintOut(LOG_INFO
, "Device: %s, NOT READY (e.g. spun down); skip device\n", device
);
2279 else if (SIMPLE_ERR_NO_MEDIUM
== err
)
2280 PrintOut(LOG_INFO
, "Device: %s, NO MEDIUM present; skip device\n", device
);
2281 else if (SIMPLE_ERR_BECOMING_READY
== err
)
2282 PrintOut(LOG_INFO
, "Device: %s, BECOMING (but not yet) READY; skip device\n", device
);
2284 PrintOut(LOG_CRIT
, "Device: %s, failed Test Unit Ready [err=%d]\n", device
, err
);
2285 CloseDevice(scsidev
, device
);
2289 // Badly-conforming USB storage devices may fail this check.
2290 // The response to the following IE mode page fetch (current and
2291 // changeable values) is carefully examined. It has been found
2292 // that various USB devices that malform the response will lock up
2293 // if asked for a log page (e.g. temperature) so it is best to
2295 if (!(err
= scsiFetchIECmpage(scsidev
, &iec
, state
.modese_len
)))
2296 state
.modese_len
= iec
.modese_len
;
2297 else if (SIMPLE_ERR_BAD_FIELD
== err
)
2298 ; /* continue since it is reasonable not to support IE mpage */
2299 else { /* any other error (including malformed response) unreasonable */
2301 "Device: %s, Bad IEC (SMART) mode page, err=%d, skip device\n",
2303 CloseDevice(scsidev
, device
);
2307 // N.B. The following is passive (i.e. it doesn't attempt to turn on
2308 // smart if it is off). This may change to be the same as the ATA side.
2309 if (!scsi_IsExceptionControlEnabled(&iec
)) {
2310 PrintOut(LOG_INFO
, "Device: %s, IE (SMART) not enabled, skip device\n"
2311 "Try 'smartctl -s on %s' to turn on SMART features\n",
2313 CloseDevice(scsidev
, device
);
2317 // Flag that certain log pages are supported (information may be
2318 // available from other sources).
2319 if (0 == scsiLogSense(scsidev
, SUPPORTED_LPAGES
, 0, tBuf
, sizeof(tBuf
), 0) ||
2320 0 == scsiLogSense(scsidev
, SUPPORTED_LPAGES
, 0, tBuf
, sizeof(tBuf
), 68))
2321 /* workaround for the bug #678 on ST8000NM0075/E001. Up to 64 pages + 4b header */
2323 for (int k
= 4; k
< tBuf
[3] + LOGPAGEHDRSIZE
; ++k
) {
2325 case TEMPERATURE_LPAGE
:
2326 state
.TempPageSupported
= 1;
2329 state
.SmartPageSupported
= 1;
2331 case READ_ERROR_COUNTER_LPAGE
:
2332 state
.ReadECounterPageSupported
= 1;
2334 case WRITE_ERROR_COUNTER_LPAGE
:
2335 state
.WriteECounterPageSupported
= 1;
2337 case VERIFY_ERROR_COUNTER_LPAGE
:
2338 state
.VerifyECounterPageSupported
= 1;
2340 case NON_MEDIUM_ERROR_LPAGE
:
2341 state
.NonMediumErrorPageSupported
= 1;
2349 // Check if scsiCheckIE() is going to work
2353 UINT8 currenttemp
= 0;
2356 if (scsiCheckIE(scsidev
, state
.SmartPageSupported
, state
.TempPageSupported
,
2357 &asc
, &ascq
, ¤ttemp
, &triptemp
)) {
2358 PrintOut(LOG_INFO
, "Device: %s, unexpectedly failed to read SMART values\n", device
);
2359 state
.SuppressReport
= 1;
2360 if (cfg
.tempdiff
|| cfg
.tempinfo
|| cfg
.tempcrit
) {
2361 PrintOut(LOG_INFO
, "Device: %s, can't monitor Temperature, ignoring -W %d,%d,%d\n",
2362 device
, cfg
.tempdiff
, cfg
.tempinfo
, cfg
.tempcrit
);
2363 cfg
.tempdiff
= cfg
.tempinfo
= cfg
.tempcrit
= 0;
2368 // capability check: self-test-log
2370 int retval
= scsiCountFailedSelfTests(scsidev
, 0);
2372 // no self-test log, turn off monitoring
2373 PrintOut(LOG_INFO
, "Device: %s, does not support SMART Self-Test Log.\n", device
);
2374 cfg
.selftest
= false;
2375 state
.selflogcount
= 0;
2376 state
.selfloghour
= 0;
2379 // register starting values to watch for changes
2380 state
.selflogcount
=SELFTEST_ERRORCOUNT(retval
);
2381 state
.selfloghour
=SELFTEST_ERRORHOURS(retval
);
2385 // disable autosave (set GLTSD bit)
2386 if (cfg
.autosave
==1){
2387 if (scsiSetControlGLTSD(scsidev
, 1, state
.modese_len
))
2388 PrintOut(LOG_INFO
,"Device: %s, could not disable autosave (set GLTSD bit).\n",device
);
2390 PrintOut(LOG_INFO
,"Device: %s, disabled autosave (set GLTSD bit).\n",device
);
2393 // or enable autosave (clear GLTSD bit)
2394 if (cfg
.autosave
==2){
2395 if (scsiSetControlGLTSD(scsidev
, 0, state
.modese_len
))
2396 PrintOut(LOG_INFO
,"Device: %s, could not enable autosave (clear GLTSD bit).\n",device
);
2398 PrintOut(LOG_INFO
,"Device: %s, enabled autosave (cleared GLTSD bit).\n",device
);
2401 // tell user we are registering device
2402 PrintOut(LOG_INFO
, "Device: %s, is SMART capable. Adding to \"monitor\" list.\n", device
);
2404 // Make sure that init_standby_check() ignores SCSI devices
2405 cfg
.offlinests_ns
= cfg
.selfteststs_ns
= false;
2407 // close file descriptor
2408 CloseDevice(scsidev
, device
);
2410 if (!state_path_prefix
.empty() || !attrlog_path_prefix
.empty()) {
2411 // Build file name for state file
2412 std::replace_if(model
, model
+strlen(model
), not_allowed_in_filename
, '_');
2413 std::replace_if(serial
, serial
+strlen(serial
), not_allowed_in_filename
, '_');
2414 if (!state_path_prefix
.empty()) {
2415 cfg
.state_file
= strprintf("%s%s-%s-%s.scsi.state", state_path_prefix
.c_str(), vendor
, model
, serial
);
2416 // Read previous state
2417 if (read_dev_state(cfg
.state_file
.c_str(), state
)) {
2418 PrintOut(LOG_INFO
, "Device: %s, state read from %s\n", device
, cfg
.state_file
.c_str());
2419 // Copy ATA attribute values to temp state
2420 state
.update_temp_state();
2423 if (!attrlog_path_prefix
.empty())
2424 cfg
.attrlog_file
= strprintf("%s%s-%s-%s.scsi.csv", attrlog_path_prefix
.c_str(), vendor
, model
, serial
);
2427 finish_device_scan(cfg
, state
);
2432 // Convert 128 bit LE integer to uint64_t or its max value on overflow.
2433 static uint64_t le128_to_uint64(const unsigned char (& val
)[16])
2435 for (int i
= 8; i
< 16; i
++) {
2437 return ~(uint64_t)0;
2439 uint64_t lo
= val
[7];
2440 for (int i
= 7-1; i
>= 0; i
--) {
2441 lo
<<= 8; lo
+= val
[i
];
2446 // Get max temperature in Kelvin reported in NVMe SMART/Health log.
2447 static int nvme_get_max_temp_kelvin(const nvme_smart_log
& smart_log
)
2449 int k
= (smart_log
.temperature
[1] << 8) | smart_log
.temperature
[0];
2450 for (int i
= 0; i
< 8; i
++) {
2451 if (smart_log
.temp_sensor
[i
] > k
)
2452 k
= smart_log
.temp_sensor
[i
];
2457 static int NVMeDeviceScan(dev_config
& cfg
, dev_state
& state
, nvme_device
* nvmedev
)
2459 const char *name
= cfg
.name
.c_str();
2461 // Device must be open
2463 // Get ID Controller
2464 nvme_id_ctrl id_ctrl
;
2465 if (!nvme_read_id_ctrl(nvmedev
, id_ctrl
)) {
2466 PrintOut(LOG_INFO
, "Device: %s, NVMe Identify Controller failed\n", name
);
2467 CloseDevice(nvmedev
, name
);
2471 // Get drive identity
2472 char model
[40+1], serial
[20+1], firmware
[8+1];
2473 format_char_array(model
, id_ctrl
.mn
);
2474 format_char_array(serial
, id_ctrl
.sn
);
2475 format_char_array(firmware
, id_ctrl
.fr
);
2477 // Format device id string for warning emails
2478 char nsstr
[32] = "", capstr
[32] = "";
2479 unsigned nsid
= nvmedev
->get_nsid();
2480 if (nsid
!= 0xffffffff)
2481 snprintf(nsstr
, sizeof(nsstr
), ", NSID:%u", nsid
);
2482 uint64_t capacity
= le128_to_uint64(id_ctrl
.tnvmcap
);
2484 format_capacity(capstr
, sizeof(capstr
), capacity
, ".");
2485 cfg
.dev_idinfo
= strprintf("%s, S/N:%s, FW:%s%s%s%s", model
, serial
, firmware
,
2486 nsstr
, (capstr
[0] ? ", " : ""), capstr
);
2488 PrintOut(LOG_INFO
, "Device: %s, %s\n", name
, cfg
.dev_idinfo
.c_str());
2490 // Read SMART/Health log
2491 nvme_smart_log smart_log
;
2492 if (!nvme_read_smart_log(nvmedev
, smart_log
)) {
2493 PrintOut(LOG_INFO
, "Device: %s, failed to read NVMe SMART/Health Information\n", name
);
2494 CloseDevice(nvmedev
, name
);
2498 // Check temperature sensor support
2499 if (cfg
.tempdiff
|| cfg
.tempinfo
|| cfg
.tempcrit
) {
2500 if (!nvme_get_max_temp_kelvin(smart_log
)) {
2501 PrintOut(LOG_INFO
, "Device: %s, no Temperature sensors, ignoring -W %d,%d,%d\n",
2502 name
, cfg
.tempdiff
, cfg
.tempinfo
, cfg
.tempcrit
);
2503 cfg
.tempdiff
= cfg
.tempinfo
= cfg
.tempcrit
= 0;
2507 // Init total error count
2508 if (cfg
.errorlog
|| cfg
.xerrorlog
) {
2509 state
.nvme_err_log_entries
= le128_to_uint64(smart_log
.num_err_log_entries
);
2512 // If no supported tests selected, return
2513 if (!( cfg
.smartcheck
|| cfg
.errorlog
|| cfg
.xerrorlog
2514 || cfg
.tempdiff
|| cfg
.tempinfo
|| cfg
.tempcrit
)) {
2515 CloseDevice(nvmedev
, name
);
2519 // Tell user we are registering device
2520 PrintOut(LOG_INFO
,"Device: %s, is SMART capable. Adding to \"monitor\" list.\n", name
);
2522 // Make sure that init_standby_check() ignores NVMe devices
2523 cfg
.offlinests_ns
= cfg
.selfteststs_ns
= false;
2525 CloseDevice(nvmedev
, name
);
2527 if (!state_path_prefix
.empty()) {
2528 // Build file name for state file
2529 std::replace_if(model
, model
+strlen(model
), not_allowed_in_filename
, '_');
2530 std::replace_if(serial
, serial
+strlen(serial
), not_allowed_in_filename
, '_');
2532 if (nsid
!= 0xffffffff)
2533 snprintf(nsstr
, sizeof(nsstr
), "-n%u", nsid
);
2534 cfg
.state_file
= strprintf("%s%s-%s%s.nvme.state", state_path_prefix
.c_str(), model
, serial
, nsstr
);
2535 // Read previous state
2536 if (read_dev_state(cfg
.state_file
.c_str(), state
))
2537 PrintOut(LOG_INFO
, "Device: %s, state read from %s\n", name
, cfg
.state_file
.c_str());
2540 finish_device_scan(cfg
, state
);
2545 // If the self-test log has got more self-test errors (or more recent
2546 // self-test errors) recorded, then notify user.
2547 static void CheckSelfTestLogs(const dev_config
& cfg
, dev_state
& state
, int newi
)
2549 const char * name
= cfg
.name
.c_str();
2553 MailWarning(cfg
, state
, 8, "Device: %s, Read SMART Self-Test Log Failed", name
);
2555 reset_warning_mail(cfg
, state
, 8, "Read SMART Self-Test Log worked again");
2557 // old and new error counts
2558 int oldc
=state
.selflogcount
;
2559 int newc
=SELFTEST_ERRORCOUNT(newi
);
2561 // old and new error timestamps in hours
2562 int oldh
=state
.selfloghour
;
2563 int newh
=SELFTEST_ERRORHOURS(newi
);
2566 // increase in error count
2567 PrintOut(LOG_CRIT
, "Device: %s, Self-Test Log error count increased from %d to %d\n",
2569 MailWarning(cfg
, state
, 3, "Device: %s, Self-Test Log error count increased from %d to %d",
2571 state
.must_write
= true;
2573 else if (newc
> 0 && oldh
!= newh
) {
2574 // more recent error
2575 // a 'more recent' error might actually be a smaller hour number,
2576 // if the hour number has wrapped.
2577 // There's still a bug here. You might just happen to run a new test
2578 // exactly 32768 hours after the previous failure, and have run exactly
2579 // 20 tests between the two, in which case smartd will miss the
2581 PrintOut(LOG_CRIT
, "Device: %s, new Self-Test Log error at hour timestamp %d\n",
2583 MailWarning(cfg
, state
, 3, "Device: %s, new Self-Test Log error at hour timestamp %d",
2585 state
.must_write
= true;
2588 // Print info if error entries have disappeared
2589 // or newer successful successful extended self-test exits
2591 PrintOut(LOG_INFO
, "Device: %s, Self-Test Log error count decreased from %d to %d\n",
2594 reset_warning_mail(cfg
, state
, 3, "Self-Test Log does no longer report errors");
2597 // Needed since self-test error count may DECREASE. Hour might
2598 // also have changed.
2599 state
.selflogcount
= newc
;
2600 state
.selfloghour
= newh
;
2605 // Test types, ordered by priority.
2606 static const char test_type_chars
[] = "LncrSCO";
2607 static const unsigned num_test_types
= sizeof(test_type_chars
)-1;
2609 // returns test type if time to do test of type testtype,
2610 // 0 if not time to do test.
2611 static char next_scheduled_test(const dev_config
& cfg
, dev_state
& state
, bool scsi
, time_t usetime
= 0)
2613 // check that self-testing has been requested
2614 if (cfg
.test_regex
.empty())
2617 // Exit if drive not capable of any test
2618 if ( state
.not_cap_long
&& state
.not_cap_short
&&
2619 (scsi
|| (state
.not_cap_conveyance
&& state
.not_cap_offline
)))
2622 // since we are about to call localtime(), be sure glibc is informed
2623 // of any timezone changes we make.
2625 FixGlibcTimeZoneBug();
2627 // Is it time for next check?
2628 time_t now
= (!usetime
? time(0) : usetime
);
2629 if (now
< state
.scheduled_test_next_check
)
2632 // Limit time check interval to 90 days
2633 if (state
.scheduled_test_next_check
+ (3600L*24*90) < now
)
2634 state
.scheduled_test_next_check
= now
- (3600L*24*90);
2636 // Check interval [state.scheduled_test_next_check, now] for scheduled tests
2638 time_t testtime
= 0; int testhour
= 0;
2639 int maxtest
= num_test_types
-1;
2641 for (time_t t
= state
.scheduled_test_next_check
; ; ) {
2642 struct tm
* tms
= localtime(&t
);
2643 // tm_wday is 0 (Sunday) to 6 (Saturday). We use 1 (Monday) to 7 (Sunday).
2644 int weekday
= (tms
->tm_wday
? tms
->tm_wday
: 7);
2645 for (int i
= 0; i
<= maxtest
; i
++) {
2646 // Skip if drive not capable of this test
2647 switch (test_type_chars
[i
]) {
2648 case 'L': if (state
.not_cap_long
) continue; break;
2649 case 'S': if (state
.not_cap_short
) continue; break;
2650 case 'C': if (scsi
|| state
.not_cap_conveyance
) continue; break;
2651 case 'O': if (scsi
|| state
.not_cap_offline
) continue; break;
2653 case 'r': if (scsi
|| state
.not_cap_selective
) continue; break;
2656 // Try match of "T/MM/DD/d/HH"
2658 snprintf(pattern
, sizeof(pattern
), "%c/%02d/%02d/%1d/%02d",
2659 test_type_chars
[i
], tms
->tm_mon
+1, tms
->tm_mday
, weekday
, tms
->tm_hour
);
2660 if (cfg
.test_regex
.full_match(pattern
)) {
2662 testtype
= pattern
[0];
2663 testtime
= t
; testhour
= tms
->tm_hour
;
2664 // Limit further matches to higher priority self-tests
2669 // Exit if no tests left or current time reached
2675 if ((t
+= 3600) > now
)
2679 // Do next check not before next hour.
2680 struct tm
* tmnow
= localtime(&now
);
2681 state
.scheduled_test_next_check
= now
+ (3600 - tmnow
->tm_min
*60 - tmnow
->tm_sec
);
2684 state
.must_write
= true;
2685 // Tell user if an old test was found.
2686 if (!usetime
&& !(testhour
== tmnow
->tm_hour
&& testtime
+ 3600 > now
)) {
2687 char datebuf
[DATEANDEPOCHLEN
]; dateandtimezoneepoch(datebuf
, testtime
);
2688 PrintOut(LOG_INFO
, "Device: %s, old test of type %c not run at %s, starting now.\n",
2689 cfg
.name
.c_str(), testtype
, datebuf
);
2696 // Print a list of future tests.
2697 static void PrintTestSchedule(const dev_config_vector
& configs
, dev_state_vector
& states
, const smart_device_list
& devices
)
2699 unsigned numdev
= configs
.size();
2702 std::vector
<int> testcnts(numdev
* num_test_types
, 0);
2704 PrintOut(LOG_INFO
, "\nNext scheduled self tests (at most 5 of each type per device):\n");
2706 // FixGlibcTimeZoneBug(); // done in PrintOut()
2707 time_t now
= time(0);
2708 char datenow
[DATEANDEPOCHLEN
], date
[DATEANDEPOCHLEN
];
2709 dateandtimezoneepoch(datenow
, now
);
2712 for (seconds
=checktime
; seconds
<3600L*24*90; seconds
+=checktime
) {
2713 // Check for each device whether a test will be run
2714 time_t testtime
= now
+ seconds
;
2715 for (unsigned i
= 0; i
< numdev
; i
++) {
2716 const dev_config
& cfg
= configs
.at(i
);
2717 dev_state
& state
= states
.at(i
);
2719 char testtype
= next_scheduled_test(cfg
, state
, devices
.at(i
)->is_scsi(), testtime
);
2720 if (testtype
&& (p
= strchr(test_type_chars
, testtype
))) {
2721 unsigned t
= (p
- test_type_chars
);
2722 // Report at most 5 tests of each type
2723 if (++testcnts
[i
*num_test_types
+ t
] <= 5) {
2724 dateandtimezoneepoch(date
, testtime
);
2725 PrintOut(LOG_INFO
, "Device: %s, will do test %d of type %c at %s\n", cfg
.name
.c_str(),
2726 testcnts
[i
*num_test_types
+ t
], testtype
, date
);
2733 dateandtimezoneepoch(date
, now
+seconds
);
2734 PrintOut(LOG_INFO
, "\nTotals [%s - %s]:\n", datenow
, date
);
2735 for (unsigned i
= 0; i
< numdev
; i
++) {
2736 const dev_config
& cfg
= configs
.at(i
);
2737 bool scsi
= devices
.at(i
)->is_scsi();
2738 for (unsigned t
= 0; t
< num_test_types
; t
++) {
2739 int cnt
= testcnts
[i
*num_test_types
+ t
];
2740 if (cnt
== 0 && !strchr((scsi
? "LS" : "LSCO"), test_type_chars
[t
]))
2742 PrintOut(LOG_INFO
, "Device: %s, will do %3d test%s of type %c\n", cfg
.name
.c_str(),
2743 cnt
, (cnt
==1?"":"s"), test_type_chars
[t
]);
2749 // Return zero on success, nonzero on failure. Perform offline (background)
2750 // short or long (extended) self test on given scsi device.
2751 static int DoSCSISelfTest(const dev_config
& cfg
, dev_state
& state
, scsi_device
* device
, char testtype
)
2754 const char *testname
= 0;
2755 const char *name
= cfg
.name
.c_str();
2758 if (scsiSelfTestInProgress(device
, &inProgress
)) {
2759 PrintOut(LOG_CRIT
, "Device: %s, does not support Self-Tests\n", name
);
2760 state
.not_cap_short
= state
.not_cap_long
= true;
2764 if (1 == inProgress
) {
2765 PrintOut(LOG_INFO
, "Device: %s, skip since Self-Test already in "
2766 "progress.\n", name
);
2772 testname
= "Short Self";
2773 retval
= scsiSmartShortSelfTest(device
);
2776 testname
= "Long Self";
2777 retval
= scsiSmartExtendSelfTest(device
);
2780 // If we can't do the test, exit
2781 if (NULL
== testname
) {
2782 PrintOut(LOG_CRIT
, "Device: %s, not capable of %c Self-Test\n", name
,
2787 if ((SIMPLE_ERR_BAD_OPCODE
== retval
) ||
2788 (SIMPLE_ERR_BAD_FIELD
== retval
)) {
2789 PrintOut(LOG_CRIT
, "Device: %s, not capable of %s-Test\n", name
,
2792 state
.not_cap_long
= true;
2794 state
.not_cap_short
= true;
2798 PrintOut(LOG_CRIT
, "Device: %s, execute %s-Test failed (err: %d)\n", name
,
2803 PrintOut(LOG_INFO
, "Device: %s, starting scheduled %s-Test.\n", name
, testname
);
2808 // Do an offline immediate or self-test. Return zero on success,
2809 // nonzero on failure.
2810 static int DoATASelfTest(const dev_config
& cfg
, dev_state
& state
, ata_device
* device
, char testtype
)
2812 const char *name
= cfg
.name
.c_str();
2814 // Read current smart data and check status/capability
2815 struct ata_smart_values data
;
2816 if (ataReadSmartValues(device
, &data
) || !(data
.offline_data_collection_capability
)) {
2817 PrintOut(LOG_CRIT
, "Device: %s, not capable of Offline or Self-Testing.\n", name
);
2821 // Check for capability to do the test
2822 int dotest
= -1, mode
= 0;
2823 const char *testname
= 0;
2826 testname
="Offline Immediate ";
2827 if (isSupportExecuteOfflineImmediate(&data
))
2828 dotest
=OFFLINE_FULL_SCAN
;
2830 state
.not_cap_offline
= true;
2833 testname
="Conveyance Self-";
2834 if (isSupportConveyanceSelfTest(&data
))
2835 dotest
=CONVEYANCE_SELF_TEST
;
2837 state
.not_cap_conveyance
= true;
2840 testname
="Short Self-";
2841 if (isSupportSelfTest(&data
))
2842 dotest
=SHORT_SELF_TEST
;
2844 state
.not_cap_short
= true;
2847 testname
="Long Self-";
2848 if (isSupportSelfTest(&data
))
2849 dotest
=EXTEND_SELF_TEST
;
2851 state
.not_cap_long
= true;
2854 case 'c': case 'n': case 'r':
2855 testname
= "Selective Self-";
2856 if (isSupportSelectiveSelfTest(&data
)) {
2857 dotest
= SELECTIVE_SELF_TEST
;
2859 case 'c': mode
= SEL_CONT
; break;
2860 case 'n': mode
= SEL_NEXT
; break;
2861 case 'r': mode
= SEL_REDO
; break;
2865 state
.not_cap_selective
= true;
2869 // If we can't do the test, exit
2871 PrintOut(LOG_CRIT
, "Device: %s, not capable of %sTest\n", name
, testname
);
2875 // If currently running a self-test, do not interrupt it to start another.
2876 if (15==(data
.self_test_exec_status
>> 4)) {
2877 if (cfg
.firmwarebugs
.is_set(BUG_SAMSUNG3
) && data
.self_test_exec_status
== 0xf0) {
2878 PrintOut(LOG_INFO
, "Device: %s, will not skip scheduled %sTest "
2879 "despite unclear Self-Test byte (SAMSUNG Firmware bug).\n", name
, testname
);
2881 PrintOut(LOG_INFO
, "Device: %s, skip scheduled %sTest; %1d0%% remaining of current Self-Test.\n",
2882 name
, testname
, (int)(data
.self_test_exec_status
& 0x0f));
2887 if (dotest
== SELECTIVE_SELF_TEST
) {
2889 ata_selective_selftest_args selargs
, prev_args
;
2890 selargs
.num_spans
= 1;
2891 selargs
.span
[0].mode
= mode
;
2892 prev_args
.num_spans
= 1;
2893 prev_args
.span
[0].start
= state
.selective_test_last_start
;
2894 prev_args
.span
[0].end
= state
.selective_test_last_end
;
2895 if (ataWriteSelectiveSelfTestLog(device
, selargs
, &data
, state
.num_sectors
, &prev_args
)) {
2896 PrintOut(LOG_CRIT
, "Device: %s, prepare %sTest failed\n", name
, testname
);
2899 uint64_t start
= selargs
.span
[0].start
, end
= selargs
.span
[0].end
;
2900 PrintOut(LOG_INFO
, "Device: %s, %s test span at LBA %" PRIu64
" - %" PRIu64
" (%" PRIu64
" sectors, %u%% - %u%% of disk).\n",
2901 name
, (selargs
.span
[0].mode
== SEL_NEXT
? "next" : "redo"),
2902 start
, end
, end
- start
+ 1,
2903 (unsigned)((100 * start
+ state
.num_sectors
/2) / state
.num_sectors
),
2904 (unsigned)((100 * end
+ state
.num_sectors
/2) / state
.num_sectors
));
2905 state
.selective_test_last_start
= start
;
2906 state
.selective_test_last_end
= end
;
2909 // execute the test, and return status
2910 int retval
= smartcommandhandler(device
, IMMEDIATE_OFFLINE
, dotest
, NULL
);
2912 PrintOut(LOG_CRIT
, "Device: %s, execute %sTest failed.\n", name
, testname
);
2916 // Report recent test start to do_disable_standby_check()
2917 // and force log of next test status
2918 if (testtype
== 'O')
2919 state
.offline_started
= true;
2921 state
.selftest_started
= true;
2923 PrintOut(LOG_INFO
, "Device: %s, starting scheduled %sTest.\n", name
, testname
);
2927 // Check pending sector count attribute values (-C, -U directives).
2928 static void check_pending(const dev_config
& cfg
, dev_state
& state
,
2929 unsigned char id
, bool increase_only
,
2930 const ata_smart_values
& smartval
,
2931 int mailtype
, const char * msg
)
2933 // Find attribute index
2934 int i
= ata_find_attr_index(id
, smartval
);
2935 if (!(i
>= 0 && ata_find_attr_index(id
, state
.smartval
) == i
))
2938 // No report if no sectors pending.
2939 uint64_t rawval
= ata_get_attr_raw_value(smartval
.vendor_attributes
[i
], cfg
.attribute_defs
);
2941 reset_warning_mail(cfg
, state
, mailtype
, "No more %s", msg
);
2945 // If attribute is not reset, report only sector count increases.
2946 uint64_t prev_rawval
= ata_get_attr_raw_value(state
.smartval
.vendor_attributes
[i
], cfg
.attribute_defs
);
2947 if (!(!increase_only
|| prev_rawval
< rawval
))
2951 std::string s
= strprintf("Device: %s, %" PRId64
" %s", cfg
.name
.c_str(), rawval
, msg
);
2952 if (prev_rawval
> 0 && rawval
!= prev_rawval
)
2953 s
+= strprintf(" (changed %+" PRId64
")", rawval
- prev_rawval
);
2955 PrintOut(LOG_CRIT
, "%s\n", s
.c_str());
2956 MailWarning(cfg
, state
, mailtype
, "%s", s
.c_str());
2957 state
.must_write
= true;
2960 // Format Temperature value
2961 static const char * fmt_temp(unsigned char x
, char (& buf
)[20])
2965 snprintf(buf
, sizeof(buf
), "%u", x
);
2969 // Check Temperature limits
2970 static void CheckTemperature(const dev_config
& cfg
, dev_state
& state
, unsigned char currtemp
, unsigned char triptemp
)
2972 if (!(0 < currtemp
&& currtemp
< 255)) {
2973 PrintOut(LOG_INFO
, "Device: %s, failed to read Temperature\n", cfg
.name
.c_str());
2977 // Update Max Temperature
2978 const char * minchg
= "", * maxchg
= "";
2979 if (currtemp
> state
.tempmax
) {
2982 state
.tempmax
= currtemp
;
2983 state
.must_write
= true;
2987 if (!state
.temperature
) {
2989 if (!state
.tempmin
|| currtemp
< state
.tempmin
)
2990 // Delay Min Temperature update by ~ 30 minutes.
2991 state
.tempmin_delay
= time(0) + CHECKTIME
- 60;
2992 PrintOut(LOG_INFO
, "Device: %s, initial Temperature is %d Celsius (Min/Max %s/%u%s)\n",
2993 cfg
.name
.c_str(), (int)currtemp
, fmt_temp(state
.tempmin
, buf
), state
.tempmax
, maxchg
);
2995 PrintOut(LOG_INFO
, " [trip Temperature is %d Celsius]\n", (int)triptemp
);
2996 state
.temperature
= currtemp
;
2999 if (state
.tempmin_delay
) {
3000 // End Min Temperature update delay if ...
3001 if ( (state
.tempmin
&& currtemp
> state
.tempmin
) // current temp exceeds recorded min,
3002 || (state
.tempmin_delay
<= time(0))) { // or delay time is over.
3003 state
.tempmin_delay
= 0;
3005 state
.tempmin
= 255;
3009 // Update Min Temperature
3010 if (!state
.tempmin_delay
&& currtemp
< state
.tempmin
) {
3011 state
.tempmin
= currtemp
;
3012 state
.must_write
= true;
3013 if (currtemp
!= state
.temperature
)
3018 if (cfg
.tempdiff
&& (*minchg
|| *maxchg
|| abs((int)currtemp
- (int)state
.temperature
) >= cfg
.tempdiff
)) {
3019 PrintOut(LOG_INFO
, "Device: %s, Temperature changed %+d Celsius to %u Celsius (Min/Max %s%s/%u%s)\n",
3020 cfg
.name
.c_str(), (int)currtemp
-(int)state
.temperature
, currtemp
, fmt_temp(state
.tempmin
, buf
), minchg
, state
.tempmax
, maxchg
);
3021 state
.temperature
= currtemp
;
3026 if (cfg
.tempcrit
&& currtemp
>= cfg
.tempcrit
) {
3027 PrintOut(LOG_CRIT
, "Device: %s, Temperature %u Celsius reached critical limit of %u Celsius (Min/Max %s%s/%u%s)\n",
3028 cfg
.name
.c_str(), currtemp
, cfg
.tempcrit
, fmt_temp(state
.tempmin
, buf
), minchg
, state
.tempmax
, maxchg
);
3029 MailWarning(cfg
, state
, 12, "Device: %s, Temperature %d Celsius reached critical limit of %u Celsius (Min/Max %s%s/%u%s)",
3030 cfg
.name
.c_str(), currtemp
, cfg
.tempcrit
, fmt_temp(state
.tempmin
, buf
), minchg
, state
.tempmax
, maxchg
);
3032 else if (cfg
.tempinfo
&& currtemp
>= cfg
.tempinfo
) {
3033 PrintOut(LOG_INFO
, "Device: %s, Temperature %u Celsius reached limit of %u Celsius (Min/Max %s%s/%u%s)\n",
3034 cfg
.name
.c_str(), currtemp
, cfg
.tempinfo
, fmt_temp(state
.tempmin
, buf
), minchg
, state
.tempmax
, maxchg
);
3036 else if (cfg
.tempcrit
) {
3037 unsigned char limit
= (cfg
.tempinfo
? cfg
.tempinfo
: cfg
.tempcrit
-5);
3038 if (currtemp
< limit
)
3039 reset_warning_mail(cfg
, state
, 12, "Temperature %u Celsius dropped below %u Celsius", currtemp
, limit
);
3043 // Check normalized and raw attribute values.
3044 static void check_attribute(const dev_config
& cfg
, dev_state
& state
,
3045 const ata_smart_attribute
& attr
,
3046 const ata_smart_attribute
& prev
,
3048 const ata_smart_threshold_entry
* thresholds
)
3050 // Check attribute and threshold
3051 ata_attr_state attrstate
= ata_get_attr_state(attr
, attridx
, thresholds
, cfg
.attribute_defs
);
3052 if (attrstate
== ATTRSTATE_NON_EXISTING
)
3055 // If requested, check for usage attributes that have failed.
3056 if ( cfg
.usagefailed
&& attrstate
== ATTRSTATE_FAILED_NOW
3057 && !cfg
.monitor_attr_flags
.is_set(attr
.id
, MONITOR_IGN_FAILUSE
)) {
3058 std::string attrname
= ata_get_smart_attr_name(attr
.id
, cfg
.attribute_defs
, cfg
.dev_rpm
);
3059 PrintOut(LOG_CRIT
, "Device: %s, Failed SMART usage Attribute: %d %s.\n", cfg
.name
.c_str(), attr
.id
, attrname
.c_str());
3060 MailWarning(cfg
, state
, 2, "Device: %s, Failed SMART usage Attribute: %d %s.", cfg
.name
.c_str(), attr
.id
, attrname
.c_str());
3061 state
.must_write
= true;
3064 // Return if we're not tracking this type of attribute
3065 bool prefail
= !!ATTRIBUTE_FLAGS_PREFAILURE(attr
.flags
);
3066 if (!( ( prefail
&& cfg
.prefail
)
3067 || (!prefail
&& cfg
.usage
)))
3070 // Return if '-I ID' was specified
3071 if (cfg
.monitor_attr_flags
.is_set(attr
.id
, MONITOR_IGNORE
))
3074 // Issue warning if they don't have the same ID in all structures.
3075 if (attr
.id
!= prev
.id
) {
3076 PrintOut(LOG_INFO
,"Device: %s, same Attribute has different ID numbers: %d = %d\n",
3077 cfg
.name
.c_str(), attr
.id
, prev
.id
);
3081 // Compare normalized values if valid.
3082 bool valchanged
= false;
3083 if (attrstate
> ATTRSTATE_NO_NORMVAL
) {
3084 if (attr
.current
!= prev
.current
)
3088 // Compare raw values if requested.
3089 bool rawchanged
= false;
3090 if (cfg
.monitor_attr_flags
.is_set(attr
.id
, MONITOR_RAW
)) {
3091 if ( ata_get_attr_raw_value(attr
, cfg
.attribute_defs
)
3092 != ata_get_attr_raw_value(prev
, cfg
.attribute_defs
))
3096 // Return if no change
3097 if (!(valchanged
|| rawchanged
))
3100 // Format value strings
3101 std::string currstr
, prevstr
;
3102 if (attrstate
== ATTRSTATE_NO_NORMVAL
) {
3103 // Print raw values only
3104 currstr
= strprintf("%s (Raw)",
3105 ata_format_attr_raw_value(attr
, cfg
.attribute_defs
).c_str());
3106 prevstr
= strprintf("%s (Raw)",
3107 ata_format_attr_raw_value(prev
, cfg
.attribute_defs
).c_str());
3109 else if (cfg
.monitor_attr_flags
.is_set(attr
.id
, MONITOR_RAW_PRINT
)) {
3110 // Print normalized and raw values
3111 currstr
= strprintf("%d [Raw %s]", attr
.current
,
3112 ata_format_attr_raw_value(attr
, cfg
.attribute_defs
).c_str());
3113 prevstr
= strprintf("%d [Raw %s]", prev
.current
,
3114 ata_format_attr_raw_value(prev
, cfg
.attribute_defs
).c_str());
3117 // Print normalized values only
3118 currstr
= strprintf("%d", attr
.current
);
3119 prevstr
= strprintf("%d", prev
.current
);
3123 std::string msg
= strprintf("Device: %s, SMART %s Attribute: %d %s changed from %s to %s",
3124 cfg
.name
.c_str(), (prefail
? "Prefailure" : "Usage"), attr
.id
,
3125 ata_get_smart_attr_name(attr
.id
, cfg
.attribute_defs
, cfg
.dev_rpm
).c_str(),
3126 prevstr
.c_str(), currstr
.c_str());
3128 // Report this change as critical ?
3129 if ( (valchanged
&& cfg
.monitor_attr_flags
.is_set(attr
.id
, MONITOR_AS_CRIT
))
3130 || (rawchanged
&& cfg
.monitor_attr_flags
.is_set(attr
.id
, MONITOR_RAW_AS_CRIT
))) {
3131 PrintOut(LOG_CRIT
, "%s\n", msg
.c_str());
3132 MailWarning(cfg
, state
, 2, "%s", msg
.c_str());
3135 PrintOut(LOG_INFO
, "%s\n", msg
.c_str());
3137 state
.must_write
= true;
3141 static int ATACheckDevice(const dev_config
& cfg
, dev_state
& state
, ata_device
* atadev
,
3142 bool firstpass
, bool allow_selftests
)
3144 const char * name
= cfg
.name
.c_str();
3146 // If user has asked, test the email warning system
3148 MailWarning(cfg
, state
, 0, "TEST EMAIL from smartd for device: %s", name
);
3150 // User may have requested (with the -n Directive) to leave the disk
3151 // alone if it is in idle or standby mode. In this case check the
3152 // power mode first before opening the device for full access,
3153 // and exit without check if disk is reported in standby.
3154 if (cfg
.powermode
&& !state
.powermodefail
) {
3155 // Note that 'is_powered_down()' handles opening the device itself, and
3156 // can be used before calling 'open()' (that's the whole point of 'is_powered_down()'!).
3157 if (atadev
->is_powered_down())
3159 // skip at most powerskipmax checks
3160 if (!cfg
.powerskipmax
|| state
.powerskipcnt
<cfg
.powerskipmax
) {
3161 // report first only except if state has changed, avoid waking up system disk
3162 if ((!state
.powerskipcnt
|| state
.lastpowermodeskipped
!= -1) && !cfg
.powerquiet
) {
3163 PrintOut(LOG_INFO
, "Device: %s, is in %s mode, suspending checks\n", name
, "STANDBY (OS)");
3164 state
.lastpowermodeskipped
= -1;
3166 state
.powerskipcnt
++;
3172 // if we can't open device, fail gracefully rather than hard --
3173 // perhaps the next time around we'll be able to open it. ATAPI
3174 // cd/dvd devices will hang awaiting media if O_NONBLOCK is not
3175 // given (see linux cdrom driver).
3176 if (!atadev
->open()) {
3177 PrintOut(LOG_INFO
, "Device: %s, open() failed: %s\n", name
, atadev
->get_errmsg());
3178 MailWarning(cfg
, state
, 9, "Device: %s, unable to open device", name
);
3182 PrintOut(LOG_INFO
,"Device: %s, opened ATA device\n", name
);
3183 reset_warning_mail(cfg
, state
, 9, "open device worked again");
3185 // user may have requested (with the -n Directive) to leave the disk
3186 // alone if it is in idle or sleeping mode. In this case check the
3187 // power mode and exit without check if needed
3188 if (cfg
.powermode
&& !state
.powermodefail
) {
3189 int dontcheck
=0, powermode
=ataCheckPowerMode(atadev
);
3190 const char * mode
= 0;
3191 if (0 <= powermode
&& powermode
< 0xff) {
3192 // wait for possible spin up and check again
3195 powermode2
= ataCheckPowerMode(atadev
);
3196 if (powermode2
> powermode
)
3197 PrintOut(LOG_INFO
, "Device: %s, CHECK POWER STATUS spins up disk (0x%02x -> 0x%02x)\n", name
, powermode
, powermode2
);
3198 powermode
= powermode2
;
3205 if (cfg
.powermode
>=1)
3211 if (cfg
.powermode
>=2)
3217 if (cfg
.powermode
>=2)
3223 if (cfg
.powermode
>=3)
3229 if (cfg
.powermode
>=3)
3235 if (cfg
.powermode
>=3)
3241 if (cfg
.powermode
>=3)
3250 mode
="ACTIVE or IDLE";
3254 PrintOut(LOG_CRIT
, "Device: %s, CHECK POWER STATUS returned %d, not ATA compliant, ignoring -n Directive\n",
3256 state
.powermodefail
= true;
3260 // if we are going to skip a check, return now
3262 // skip at most powerskipmax checks
3263 if (!cfg
.powerskipmax
|| state
.powerskipcnt
<cfg
.powerskipmax
) {
3264 CloseDevice(atadev
, name
);
3265 // report first only except if state has changed, avoid waking up system disk
3266 if ((!state
.powerskipcnt
|| state
.lastpowermodeskipped
!= powermode
) && !cfg
.powerquiet
) {
3267 PrintOut(LOG_INFO
, "Device: %s, is in %s mode, suspending checks\n", name
, mode
);
3268 state
.lastpowermodeskipped
= powermode
;
3270 state
.powerskipcnt
++;
3274 PrintOut(LOG_INFO
, "Device: %s, %s mode ignored due to reached limit of skipped checks (%d check%s skipped)\n",
3275 name
, mode
, state
.powerskipcnt
, (state
.powerskipcnt
==1?"":"s"));
3277 state
.powerskipcnt
= 0;
3278 state
.tempmin_delay
= time(0) + CHECKTIME
- 60; // Delay Min Temperature update
3280 else if (state
.powerskipcnt
) {
3281 PrintOut(LOG_INFO
, "Device: %s, is back in %s mode, resuming checks (%d check%s skipped)\n",
3282 name
, mode
, state
.powerskipcnt
, (state
.powerskipcnt
==1?"":"s"));
3283 state
.powerskipcnt
= 0;
3284 state
.tempmin_delay
= time(0) + CHECKTIME
- 60; // Delay Min Temperature update
3288 // check smart status
3289 if (cfg
.smartcheck
) {
3290 int status
=ataSmartStatus2(atadev
);
3292 PrintOut(LOG_INFO
,"Device: %s, not capable of SMART self-check\n",name
);
3293 MailWarning(cfg
, state
, 5, "Device: %s, not capable of SMART self-check", name
);
3294 state
.must_write
= true;
3296 else if (status
==1){
3297 PrintOut(LOG_CRIT
, "Device: %s, FAILED SMART self-check. BACK UP DATA NOW!\n", name
);
3298 MailWarning(cfg
, state
, 1, "Device: %s, FAILED SMART self-check. BACK UP DATA NOW!", name
);
3299 state
.must_write
= true;
3303 // Check everything that depends upon SMART Data (eg, Attribute values)
3304 if ( cfg
.usagefailed
|| cfg
.prefail
|| cfg
.usage
3305 || cfg
.curr_pending_id
|| cfg
.offl_pending_id
3306 || cfg
.tempdiff
|| cfg
.tempinfo
|| cfg
.tempcrit
3307 || cfg
.selftest
|| cfg
.offlinests
|| cfg
.selfteststs
) {
3309 // Read current attribute values.
3310 ata_smart_values curval
;
3311 if (ataReadSmartValues(atadev
, &curval
)){
3312 PrintOut(LOG_CRIT
, "Device: %s, failed to read SMART Attribute Data\n", name
);
3313 MailWarning(cfg
, state
, 6, "Device: %s, failed to read SMART Attribute Data", name
);
3314 state
.must_write
= true;
3317 reset_warning_mail(cfg
, state
, 6, "read SMART Attribute Data worked again");
3319 // look for current or offline pending sectors
3320 if (cfg
.curr_pending_id
)
3321 check_pending(cfg
, state
, cfg
.curr_pending_id
, cfg
.curr_pending_incr
, curval
, 10,
3322 (!cfg
.curr_pending_incr
? "Currently unreadable (pending) sectors"
3323 : "Total unreadable (pending) sectors" ));
3325 if (cfg
.offl_pending_id
)
3326 check_pending(cfg
, state
, cfg
.offl_pending_id
, cfg
.offl_pending_incr
, curval
, 11,
3327 (!cfg
.offl_pending_incr
? "Offline uncorrectable sectors"
3328 : "Total offline uncorrectable sectors"));
3330 // check temperature limits
3331 if (cfg
.tempdiff
|| cfg
.tempinfo
|| cfg
.tempcrit
)
3332 CheckTemperature(cfg
, state
, ata_return_temperature_value(&curval
, cfg
.attribute_defs
), 0);
3334 // look for failed usage attributes, or track usage or prefail attributes
3335 if (cfg
.usagefailed
|| cfg
.prefail
|| cfg
.usage
) {
3336 for (int i
= 0; i
< NUMBER_ATA_SMART_ATTRIBUTES
; i
++) {
3337 check_attribute(cfg
, state
,
3338 curval
.vendor_attributes
[i
],
3339 state
.smartval
.vendor_attributes
[i
],
3340 i
, state
.smartthres
.thres_entries
);
3344 // Log changes of offline data collection status
3345 if (cfg
.offlinests
) {
3346 if ( curval
.offline_data_collection_status
3347 != state
.smartval
.offline_data_collection_status
3348 || state
.offline_started
// test was started in previous call
3349 || (firstpass
&& (debugmode
|| (curval
.offline_data_collection_status
& 0x7d))))
3350 log_offline_data_coll_status(name
, curval
.offline_data_collection_status
);
3353 // Log changes of self-test execution status
3354 if (cfg
.selfteststs
) {
3355 if ( curval
.self_test_exec_status
!= state
.smartval
.self_test_exec_status
3356 || state
.selftest_started
// test was started in previous call
3357 || (firstpass
&& (debugmode
|| curval
.self_test_exec_status
!= 0x00)))
3358 log_self_test_exec_status(name
, curval
.self_test_exec_status
);
3361 // Save the new values for the next time around
3362 state
.smartval
= curval
;
3365 state
.offline_started
= state
.selftest_started
= false;
3367 // check if number of selftest errors has increased (note: may also DECREASE)
3369 CheckSelfTestLogs(cfg
, state
, SelfTestErrorCount(atadev
, name
, cfg
.firmwarebugs
));
3371 // check if number of ATA errors has increased
3372 if (cfg
.errorlog
|| cfg
.xerrorlog
) {
3374 int errcnt1
= -1, errcnt2
= -1;
3376 errcnt1
= read_ata_error_count(atadev
, name
, cfg
.firmwarebugs
, false);
3378 errcnt2
= read_ata_error_count(atadev
, name
, cfg
.firmwarebugs
, true);
3380 // new number of errors is max of both logs
3381 int newc
= (errcnt1
>= errcnt2
? errcnt1
: errcnt2
);
3383 // did command fail?
3385 // lack of PrintOut here is INTENTIONAL
3386 MailWarning(cfg
, state
, 7, "Device: %s, Read SMART Error Log Failed", name
);
3388 // has error count increased?
3389 int oldc
= state
.ataerrorcount
;
3391 PrintOut(LOG_CRIT
, "Device: %s, ATA error count increased from %d to %d\n",
3393 MailWarning(cfg
, state
, 4, "Device: %s, ATA error count increased from %d to %d",
3395 state
.must_write
= true;
3399 state
.ataerrorcount
=newc
;
3402 // if the user has asked, and device is capable (or we're not yet
3403 // sure) check whether a self test should be done now.
3404 if (allow_selftests
&& !cfg
.test_regex
.empty()) {
3405 char testtype
= next_scheduled_test(cfg
, state
, false/*!scsi*/);
3407 DoATASelfTest(cfg
, state
, atadev
, testtype
);
3410 // Don't leave device open -- the OS/user may want to access it
3411 // before the next smartd cycle!
3412 CloseDevice(atadev
, name
);
3414 // Copy ATA attribute values to persistent state
3415 state
.update_persistent_state();
3420 static int SCSICheckDevice(const dev_config
& cfg
, dev_state
& state
, scsi_device
* scsidev
, bool allow_selftests
)
3422 const char * name
= cfg
.name
.c_str();
3424 // If the user has asked for it, test the email warning system
3426 MailWarning(cfg
, state
, 0, "TEST EMAIL from smartd for device: %s", name
);
3428 // if we can't open device, fail gracefully rather than hard --
3429 // perhaps the next time around we'll be able to open it
3430 if (!scsidev
->open()) {
3431 PrintOut(LOG_INFO
, "Device: %s, open() failed: %s\n", name
, scsidev
->get_errmsg());
3432 MailWarning(cfg
, state
, 9, "Device: %s, unable to open device", name
);
3434 } else if (debugmode
)
3435 PrintOut(LOG_INFO
,"Device: %s, opened SCSI device\n", name
);
3436 reset_warning_mail(cfg
, state
, 9, "open device worked again");
3438 UINT8 asc
= 0, ascq
= 0;
3439 UINT8 currenttemp
= 0, triptemp
= 0;
3440 if (!state
.SuppressReport
) {
3441 if (scsiCheckIE(scsidev
, state
.SmartPageSupported
, state
.TempPageSupported
,
3442 &asc
, &ascq
, ¤ttemp
, &triptemp
)) {
3443 PrintOut(LOG_INFO
, "Device: %s, failed to read SMART values\n",
3445 MailWarning(cfg
, state
, 6, "Device: %s, failed to read SMART values", name
);
3446 state
.SuppressReport
= 1;
3450 const char * cp
= scsiGetIEString(asc
, ascq
);
3452 PrintOut(LOG_CRIT
, "Device: %s, SMART Failure: %s\n", name
, cp
);
3453 MailWarning(cfg
, state
, 1,"Device: %s, SMART Failure: %s", name
, cp
);
3454 } else if (asc
== 4 && ascq
== 9) {
3455 PrintOut(LOG_INFO
,"Device: %s, self-test in progress\n", name
);
3456 } else if (debugmode
)
3457 PrintOut(LOG_INFO
,"Device: %s, non-SMART asc,ascq: %d,%d\n",
3458 name
, (int)asc
, (int)ascq
);
3459 } else if (debugmode
)
3460 PrintOut(LOG_INFO
,"Device: %s, SMART health: passed\n", name
);
3462 // check temperature limits
3463 if (cfg
.tempdiff
|| cfg
.tempinfo
|| cfg
.tempcrit
|| !cfg
.attrlog_file
.empty())
3464 CheckTemperature(cfg
, state
, currenttemp
, triptemp
);
3466 // check if number of selftest errors has increased (note: may also DECREASE)
3468 CheckSelfTestLogs(cfg
, state
, scsiCountFailedSelfTests(scsidev
, 0));
3470 if (allow_selftests
&& !cfg
.test_regex
.empty()) {
3471 char testtype
= next_scheduled_test(cfg
, state
, true/*scsi*/);
3473 DoSCSISelfTest(cfg
, state
, scsidev
, testtype
);
3475 if (!cfg
.attrlog_file
.empty()){
3476 // saving error counters to state
3478 if (state
.ReadECounterPageSupported
&& (0 == scsiLogSense(scsidev
,
3479 READ_ERROR_COUNTER_LPAGE
, 0, tBuf
, sizeof(tBuf
), 0))) {
3480 scsiDecodeErrCounterPage(tBuf
, &state
.scsi_error_counters
[0].errCounter
);
3481 state
.scsi_error_counters
[0].found
=1;
3483 if (state
.WriteECounterPageSupported
&& (0 == scsiLogSense(scsidev
,
3484 WRITE_ERROR_COUNTER_LPAGE
, 0, tBuf
, sizeof(tBuf
), 0))) {
3485 scsiDecodeErrCounterPage(tBuf
, &state
.scsi_error_counters
[1].errCounter
);
3486 state
.scsi_error_counters
[1].found
=1;
3488 if (state
.VerifyECounterPageSupported
&& (0 == scsiLogSense(scsidev
,
3489 VERIFY_ERROR_COUNTER_LPAGE
, 0, tBuf
, sizeof(tBuf
), 0))) {
3490 scsiDecodeErrCounterPage(tBuf
, &state
.scsi_error_counters
[2].errCounter
);
3491 state
.scsi_error_counters
[2].found
=1;
3493 if (state
.NonMediumErrorPageSupported
&& (0 == scsiLogSense(scsidev
,
3494 NON_MEDIUM_ERROR_LPAGE
, 0, tBuf
, sizeof(tBuf
), 0))) {
3495 scsiDecodeNonMediumErrPage(tBuf
, &state
.scsi_nonmedium_error
.nme
);
3496 state
.scsi_nonmedium_error
.found
=1;
3499 CloseDevice(scsidev
, name
);
3503 static int NVMeCheckDevice(const dev_config
& cfg
, dev_state
& state
, nvme_device
* nvmedev
)
3505 const char * name
= cfg
.name
.c_str();
3507 // TODO: Use common open function for ATA/SCSI/NVMe
3508 // If user has asked, test the email warning system
3510 MailWarning(cfg
, state
, 0, "TEST EMAIL from smartd for device: %s", name
);
3512 if (!nvmedev
->open()) {
3513 PrintOut(LOG_INFO
, "Device: %s, open() failed: %s\n", name
, nvmedev
->get_errmsg());
3514 MailWarning(cfg
, state
, 9, "Device: %s, unable to open device", name
);
3518 PrintOut(LOG_INFO
,"Device: %s, opened NVMe device\n", name
);
3519 reset_warning_mail(cfg
, state
, 9, "open device worked again");
3521 // Read SMART/Health log
3522 nvme_smart_log smart_log
;
3523 if (!nvme_read_smart_log(nvmedev
, smart_log
)) {
3524 PrintOut(LOG_INFO
, "Device: %s, failed to read NVMe SMART/Health Information\n", name
);
3525 MailWarning(cfg
, state
, 6, "Device: %s, failed to read NVMe SMART/Health Information", name
);
3526 state
.must_write
= true;
3530 // Check Critical Warning bits
3531 if (cfg
.smartcheck
&& smart_log
.critical_warning
) {
3532 unsigned char w
= smart_log
.critical_warning
;
3534 static const char * const wnames
[] =
3535 {"LowSpare", "Temperature", "Reliability", "R/O", "VolMemBackup"};
3537 for (unsigned b
= 0, cnt
= 0; b
< 8 ; b
++) {
3538 if (!(w
& (1 << b
)))
3543 msg
+= "..."; break;
3545 if (b
>= sizeof(wnames
)/sizeof(wnames
[0])) {
3546 msg
+= "*Unknown*"; break;
3551 PrintOut(LOG_CRIT
, "Device: %s, Critical Warning (0x%02x): %s\n", name
, w
, msg
.c_str());
3552 MailWarning(cfg
, state
, 1, "Device: %s, Critical Warning (0x%02x): %s", name
, w
, msg
.c_str());
3553 state
.must_write
= true;
3556 // Check temperature limits
3557 if (cfg
.tempdiff
|| cfg
.tempinfo
|| cfg
.tempcrit
) {
3558 int k
= nvme_get_max_temp_kelvin(smart_log
);
3559 // Convert Kelvin to positive Celsius (TODO: Allow negative temperatures)
3565 CheckTemperature(cfg
, state
, c
, 0);
3568 // Check if number of errors has increased
3569 if (cfg
.errorlog
|| cfg
.xerrorlog
) {
3570 uint64_t oldcnt
= state
.nvme_err_log_entries
;
3571 uint64_t newcnt
= le128_to_uint64(smart_log
.num_err_log_entries
);
3572 if (newcnt
> oldcnt
) {
3573 PrintOut(LOG_CRIT
, "Device: %s, number of Error Log entries increased from %" PRIu64
" to %" PRIu64
"\n",
3574 name
, oldcnt
, newcnt
);
3575 MailWarning(cfg
, state
, 4, "Device: %s, number of Error Log entries increased from %" PRIu64
" to %" PRIu64
,
3576 name
, oldcnt
, newcnt
);
3577 state
.must_write
= true;
3579 state
.nvme_err_log_entries
= newcnt
;
3582 CloseDevice(nvmedev
, name
);
3586 // 0=not used, 1=not disabled, 2=disable rejected by OS, 3=disabled
3587 static int standby_disable_state
= 0;
3589 static void init_disable_standby_check(dev_config_vector
& configs
)
3591 // Check for '-l offlinests,ns' or '-l selfteststs,ns' directives
3592 bool sts1
= false, sts2
= false;
3593 for (unsigned i
= 0; i
< configs
.size() && !(sts1
|| sts2
); i
++) {
3594 const dev_config
& cfg
= configs
.at(i
);
3595 if (cfg
.offlinests_ns
)
3597 if (cfg
.selfteststs_ns
)
3601 // Check for support of disable auto standby
3602 // Reenable standby if smartd.conf was reread
3603 if (sts1
|| sts2
|| standby_disable_state
== 3) {
3604 if (!smi()->disable_system_auto_standby(false)) {
3605 if (standby_disable_state
== 3)
3606 PrintOut(LOG_CRIT
, "System auto standby enable failed: %s\n", smi()->get_errmsg());
3608 PrintOut(LOG_INFO
, "Disable auto standby not supported, ignoring ',ns' from %s%s%s\n",
3609 (sts1
? "-l offlinests,ns" : ""), (sts1
&& sts2
? " and " : ""), (sts2
? "-l selfteststs,ns" : ""));
3610 sts1
= sts2
= false;
3615 standby_disable_state
= (sts1
|| sts2
? 1 : 0);
3618 static void do_disable_standby_check(const dev_config_vector
& configs
, const dev_state_vector
& states
)
3620 if (!standby_disable_state
)
3623 // Check for just started or still running self-tests
3624 bool running
= false;
3625 for (unsigned i
= 0; i
< configs
.size() && !running
; i
++) {
3626 const dev_config
& cfg
= configs
.at(i
); const dev_state
& state
= states
.at(i
);
3628 if ( ( cfg
.offlinests_ns
3629 && (state
.offline_started
||
3630 is_offl_coll_in_progress(state
.smartval
.offline_data_collection_status
)))
3631 || ( cfg
.selfteststs_ns
3632 && (state
.selftest_started
||
3633 is_self_test_in_progress(state
.smartval
.self_test_exec_status
))) )
3635 // state.offline/selftest_started will be reset after next logging of test status
3638 // Disable/enable auto standby and log state changes
3640 if (standby_disable_state
!= 1) {
3641 if (!smi()->disable_system_auto_standby(false))
3642 PrintOut(LOG_CRIT
, "Self-test(s) completed, system auto standby enable failed: %s\n",
3643 smi()->get_errmsg());
3645 PrintOut(LOG_INFO
, "Self-test(s) completed, system auto standby enabled\n");
3646 standby_disable_state
= 1;
3649 else if (!smi()->disable_system_auto_standby(true)) {
3650 if (standby_disable_state
!= 2) {
3651 PrintOut(LOG_INFO
, "Self-test(s) in progress, system auto standby disable rejected: %s\n",
3652 smi()->get_errmsg());
3653 standby_disable_state
= 2;
3657 if (standby_disable_state
!= 3) {
3658 PrintOut(LOG_INFO
, "Self-test(s) in progress, system auto standby disabled\n");
3659 standby_disable_state
= 3;
3664 // Checks the SMART status of all ATA and SCSI devices
3665 static void CheckDevicesOnce(const dev_config_vector
& configs
, dev_state_vector
& states
,
3666 smart_device_list
& devices
, bool firstpass
, bool allow_selftests
)
3668 for (unsigned i
= 0; i
< configs
.size(); i
++) {
3669 const dev_config
& cfg
= configs
.at(i
);
3670 dev_state
& state
= states
.at(i
);
3671 smart_device
* dev
= devices
.at(i
);
3673 ATACheckDevice(cfg
, state
, dev
->to_ata(), firstpass
, allow_selftests
);
3674 else if (dev
->is_scsi())
3675 SCSICheckDevice(cfg
, state
, dev
->to_scsi(), allow_selftests
);
3676 else if (dev
->is_nvme())
3677 NVMeCheckDevice(cfg
, state
, dev
->to_nvme());
3680 do_disable_standby_check(configs
, states
);
3683 // Set if Initialize() was called
3684 static bool is_initialized
= false;
3686 // Does initialization right after fork to daemon mode
3687 static void Initialize(time_t *wakeuptime
)
3689 // Call Goodbye() on exit
3690 is_initialized
= true;
3696 // install signal handlers. On Solaris, can't use signal() because
3697 // it resets the handler to SIG_DFL after each call. So use sigset()
3698 // instead. So SIGNALFN()==signal() or SIGNALFN()==sigset().
3700 // normal and abnormal exit
3701 if (SIGNALFN(SIGTERM
, sighandler
)==SIG_IGN
)
3702 SIGNALFN(SIGTERM
, SIG_IGN
);
3703 if (SIGNALFN(SIGQUIT
, sighandler
)==SIG_IGN
)
3704 SIGNALFN(SIGQUIT
, SIG_IGN
);
3706 // in debug mode, <CONTROL-C> ==> HUP
3707 if (SIGNALFN(SIGINT
, debugmode
?HUPhandler
:sighandler
)==SIG_IGN
)
3708 SIGNALFN(SIGINT
, SIG_IGN
);
3710 // Catch HUP and USR1
3711 if (SIGNALFN(SIGHUP
, HUPhandler
)==SIG_IGN
)
3712 SIGNALFN(SIGHUP
, SIG_IGN
);
3713 if (SIGNALFN(SIGUSR1
, USR1handler
)==SIG_IGN
)
3714 SIGNALFN(SIGUSR1
, SIG_IGN
);
3716 if (SIGNALFN(SIGUSR2
, USR2handler
)==SIG_IGN
)
3717 SIGNALFN(SIGUSR2
, SIG_IGN
);
3720 // initialize wakeup time to CURRENT time
3721 *wakeuptime
=time(NULL
);
3727 // Toggle debug mode implemented for native windows only
3728 // (there is no easy way to reopen tty on *nix)
3729 static void ToggleDebugMode()
3732 PrintOut(LOG_INFO
,"Signal USR2 - enabling debug mode\n");
3733 if (!daemon_enable_console("smartd [Debug]")) {
3735 daemon_signal(SIGINT
, HUPhandler
);
3736 PrintOut(LOG_INFO
,"smartd debug mode enabled, PID=%d\n", getpid());
3739 PrintOut(LOG_INFO
,"enable console failed\n");
3741 else if (debugmode
== 1) {
3742 daemon_disable_console();
3744 daemon_signal(SIGINT
, sighandler
);
3745 PrintOut(LOG_INFO
,"Signal USR2 - debug mode disabled\n");
3748 PrintOut(LOG_INFO
,"Signal USR2 - debug mode %d not changed\n", debugmode
);
3752 static time_t dosleep(time_t wakeuptime
, bool & sigwakeup
)
3754 // If past wake-up-time, compute next wake-up-time
3755 time_t timenow
=time(NULL
);
3756 while (wakeuptime
<=timenow
){
3757 int intervals
=1+(timenow
-wakeuptime
)/checktime
;
3758 wakeuptime
+=intervals
*checktime
;
3761 // sleep until we catch SIGUSR1 or have completed sleeping
3763 while (timenow
< wakeuptime
+addtime
&& !caughtsigUSR1
&& !caughtsigHUP
&& !caughtsigEXIT
) {
3765 // protect user again system clock being adjusted backwards
3766 if (wakeuptime
>timenow
+checktime
){
3767 PrintOut(LOG_CRIT
, "System clock time adjusted to the past. Resetting next wakeup time.\n");
3768 wakeuptime
=timenow
+checktime
;
3771 // Exit sleep when time interval has expired or a signal is received
3772 sleep(wakeuptime
+addtime
-timenow
);
3775 // toggle debug mode?
3776 if (caughtsigUSR2
) {
3784 // Actual sleep time too long?
3785 if (!addtime
&& timenow
> wakeuptime
+60) {
3787 PrintOut(LOG_INFO
, "Sleep time was %d seconds too long, assuming wakeup from standby mode.\n",
3788 (int)(timenow
-wakeuptime
));
3789 // Wait another 20 seconds to avoid I/O errors during disk spin-up
3790 addtime
= timenow
-wakeuptime
+20;
3791 // Use next wake-up-time if close
3792 int nextcheck
= checktime
- addtime
% checktime
;
3793 if (nextcheck
<= 20)
3794 addtime
+= nextcheck
;
3798 // if we caught a SIGUSR1 then print message and clear signal
3800 PrintOut(LOG_INFO
,"Signal USR1 - checking devices now rather than in %d seconds.\n",
3801 wakeuptime
-timenow
>0?(int)(wakeuptime
-timenow
):0);
3806 // return adjusted wakeuptime
3810 // Print out a list of valid arguments for the Directive d
3811 static void printoutvaliddirectiveargs(int priority
, char d
)
3815 PrintOut(priority
, "never[,N][,q], sleep[,N][,q], standby[,N][,q], idle[,N][,q]");
3818 PrintOut(priority
, "valid_regular_expression");
3821 PrintOut(priority
, "%s", smi()->get_valid_dev_types_str().c_str());
3824 PrintOut(priority
, "normal, permissive");
3828 PrintOut(priority
, "on, off");
3831 PrintOut(priority
, "error, selftest");
3834 PrintOut(priority
, "\"once\", \"daily\", \"diminishing\", \"test\", \"exec\"");
3837 PrintOut(priority
, "\n%s\n", create_vendor_attribute_arg_list().c_str());
3840 PrintOut(priority
, "use, ignore, show, showall");
3843 PrintOut(priority
, "%s", get_valid_firmwarebug_args());
3846 PrintOut(priority
, "aam,[N|off], apm,[N|off], lookahead,[on|off], "
3847 "security-freeze, standby,[N|off], wcache,[on|off]");
3852 // exits with an error message, or returns integer value of token
3853 static int GetInteger(const char *arg
, const char *name
, const char *token
, int lineno
, const char *cfgfile
,
3854 int min
, int max
, char * suffix
= 0)
3856 // make sure argument is there
3858 PrintOut(LOG_CRIT
,"File %s line %d (drive %s): Directive: %s takes integer argument from %d to %d.\n",
3859 cfgfile
, lineno
, name
, token
, min
, max
);
3863 // get argument value (base 10), check that it's integer, and in-range
3865 int val
= strtol(arg
,&endptr
,10);
3867 // optional suffix present?
3869 if (!strcmp(endptr
, suffix
))
3870 endptr
+= strlen(suffix
);
3875 if (!(!*endptr
&& min
<= val
&& val
<= max
)) {
3876 PrintOut(LOG_CRIT
,"File %s line %d (drive %s): Directive: %s has argument: %s; needs integer from %d to %d.\n",
3877 cfgfile
, lineno
, name
, token
, arg
, min
, max
);
3881 // all is well; return value
3886 // Get 1-3 small integer(s) for '-W' directive
3887 static int Get3Integers(const char *arg
, const char *name
, const char *token
, int lineno
, const char *cfgfile
,
3888 unsigned char *val1
, unsigned char *val2
, unsigned char *val3
)
3890 unsigned v1
= 0, v2
= 0, v3
= 0;
3891 int n1
= -1, n2
= -1, n3
= -1, len
;
3893 PrintOut(LOG_CRIT
,"File %s line %d (drive %s): Directive: %s takes 1-3 integer argument(s) from 0 to 255.\n",
3894 cfgfile
, lineno
, name
, token
);
3899 if (!( sscanf(arg
, "%u%n,%u%n,%u%n", &v1
, &n1
, &v2
, &n2
, &v3
, &n3
) >= 1
3900 && (n1
== len
|| n2
== len
|| n3
== len
) && v1
<= 255 && v2
<= 255 && v3
<= 255)) {
3901 PrintOut(LOG_CRIT
,"File %s line %d (drive %s): Directive: %s has argument: %s; needs 1-3 integer(s) from 0 to 255.\n",
3902 cfgfile
, lineno
, name
, token
, arg
);
3905 *val1
= (unsigned char)v1
; *val2
= (unsigned char)v2
; *val3
= (unsigned char)v3
;
3912 // Concatenate strtok() results if quoted with "..."
3913 static const char * strtok_dequote(const char * delimiters
)
3915 const char * t
= strtok(0, delimiters
);
3916 if (!t
|| t
[0] != '"')
3919 static std::string token
;
3922 t
= strtok(0, delimiters
);
3926 int len
= strlen(t
);
3927 if (t
[len
-1] == '"') {
3928 token
+= std::string(t
, len
-1);
3933 return token
.c_str();
3939 // This function returns 1 if it has correctly parsed one token (and
3940 // any arguments), else zero if no tokens remain. It returns -1 if an
3941 // error was encountered.
3942 static int ParseToken(char * token
, dev_config
& cfg
, smart_devtype_list
& scan_types
)
3945 const char * name
= cfg
.name
.c_str();
3946 int lineno
=cfg
.lineno
;
3947 const char *delim
= " \n\t";
3950 const char *arg
= 0;
3952 // is the rest of the line a comment
3956 // is the token not recognized?
3957 if (*token
!='-' || strlen(token
)!=2) {
3958 PrintOut(LOG_CRIT
,"File %s line %d (drive %s): unknown Directive: %s\n",
3959 configfile
, lineno
, name
, token
);
3960 PrintOut(LOG_CRIT
, "Run smartd -D to print a list of valid Directives.\n");
3964 // token we will be parsing:
3967 // parse the token and swallow its argument
3969 char plus
[] = "+", excl
[] = "!";
3973 // monitor current pending sector count (default 197)
3974 if ((val
= GetInteger(arg
=strtok(NULL
,delim
), name
, token
, lineno
, configfile
, 0, 255, plus
)) < 0)
3976 cfg
.curr_pending_id
= (unsigned char)val
;
3977 cfg
.curr_pending_incr
= (*plus
== '+');
3978 cfg
.curr_pending_set
= true;
3981 // monitor offline uncorrectable sectors (default 198)
3982 if ((val
= GetInteger(arg
=strtok(NULL
,delim
), name
, token
, lineno
, configfile
, 0, 255, plus
)) < 0)
3984 cfg
.offl_pending_id
= (unsigned char)val
;
3985 cfg
.offl_pending_incr
= (*plus
== '+');
3986 cfg
.offl_pending_set
= true;
3989 // Set tolerance level for SMART command failures
3990 if ((arg
= strtok(NULL
, delim
)) == NULL
) {
3992 } else if (!strcmp(arg
, "normal")) {
3993 // Normal mode: exit on failure of a mandatory S.M.A.R.T. command, but
3994 // not on failure of an optional S.M.A.R.T. command.
3995 // This is the default so we don't need to actually do anything here.
3996 cfg
.permissive
= false;
3997 } else if (!strcmp(arg
, "permissive")) {
3998 // Permissive mode; ignore errors from Mandatory SMART commands
3999 cfg
.permissive
= true;
4005 // specify the device type
4006 if ((arg
= strtok(NULL
, delim
)) == NULL
) {
4008 } else if (!strcmp(arg
, "ignore")) {
4010 } else if (!strcmp(arg
, "removable")) {
4011 cfg
.removable
= true;
4012 } else if (!strcmp(arg
, "auto")) {
4017 scan_types
.push_back(arg
);
4022 if (!(arg
= strtok(0, delim
)))
4024 else if (!parse_firmwarebug_def(arg
, cfg
.firmwarebugs
))
4028 // check SMART status
4029 cfg
.smartcheck
= true;
4032 // check for failure of usage attributes
4033 cfg
.usagefailed
= true;
4036 // track changes in all vendor attributes
4041 // track changes in prefail vendor attributes
4045 // track changes in usage vendor attributes
4049 // track changes in SMART logs
4050 if ((arg
= strtok(NULL
, delim
)) == NULL
) {
4052 } else if (!strcmp(arg
, "selftest")) {
4053 // track changes in self-test log
4054 cfg
.selftest
= true;
4055 } else if (!strcmp(arg
, "error")) {
4056 // track changes in ATA error log
4057 cfg
.errorlog
= true;
4058 } else if (!strcmp(arg
, "xerror")) {
4059 // track changes in Extended Comprehensive SMART error log
4060 cfg
.xerrorlog
= true;
4061 } else if (!strcmp(arg
, "offlinests")) {
4062 // track changes in offline data collection status
4063 cfg
.offlinests
= true;
4064 } else if (!strcmp(arg
, "offlinests,ns")) {
4065 // track changes in offline data collection status, disable auto standby
4066 cfg
.offlinests
= cfg
.offlinests_ns
= true;
4067 } else if (!strcmp(arg
, "selfteststs")) {
4068 // track changes in self-test execution status
4069 cfg
.selfteststs
= true;
4070 } else if (!strcmp(arg
, "selfteststs,ns")) {
4071 // track changes in self-test execution status, disable auto standby
4072 cfg
.selfteststs
= cfg
.selfteststs_ns
= true;
4073 } else if (!strncmp(arg
, "scterc,", sizeof("scterc,")-1)) {
4074 // set SCT Error Recovery Control
4075 unsigned rt
= ~0, wt
= ~0; int nc
= -1;
4076 sscanf(arg
,"scterc,%u,%u%n", &rt
, &wt
, &nc
);
4077 if (nc
== (int)strlen(arg
) && rt
<= 999 && wt
<= 999) {
4078 cfg
.sct_erc_set
= true;
4079 cfg
.sct_erc_readtime
= rt
;
4080 cfg
.sct_erc_writetime
= wt
;
4089 // monitor everything
4090 cfg
.smartcheck
= true;
4092 cfg
.usagefailed
= true;
4094 cfg
.selftest
= true;
4095 cfg
.errorlog
= true;
4096 cfg
.selfteststs
= true;
4099 // automatic offline testing enable/disable
4100 if ((arg
= strtok(NULL
, delim
)) == NULL
) {
4102 } else if (!strcmp(arg
, "on")) {
4103 cfg
.autoofflinetest
= 2;
4104 } else if (!strcmp(arg
, "off")) {
4105 cfg
.autoofflinetest
= 1;
4111 // skip disk check if in idle or standby mode
4112 if (!(arg
= strtok(NULL
, delim
)))
4115 char *endptr
= NULL
;
4116 char *next
= strchr(const_cast<char*>(arg
), ',');
4118 cfg
.powerquiet
= false;
4119 cfg
.powerskipmax
= 0;
4121 if (next
!=NULL
) *next
='\0';
4122 if (!strcmp(arg
, "never"))
4124 else if (!strcmp(arg
, "sleep"))
4126 else if (!strcmp(arg
, "standby"))
4128 else if (!strcmp(arg
, "idle"))
4133 // if optional arguments are present
4134 if (!badarg
&& next
!=NULL
) {
4136 cfg
.powerskipmax
= strtol(next
, &endptr
, 10);
4138 cfg
.powerskipmax
= 0;
4140 next
= endptr
+ (*endptr
!= '\0');
4141 if (cfg
.powerskipmax
<= 0)
4144 if (*next
!= '\0') {
4145 if (!strcmp("q", next
))
4146 cfg
.powerquiet
= true;
4155 // automatic attribute autosave enable/disable
4156 if ((arg
= strtok(NULL
, delim
)) == NULL
) {
4158 } else if (!strcmp(arg
, "on")) {
4160 } else if (!strcmp(arg
, "off")) {
4167 // warn user, and delete any previously given -s REGEXP Directives
4168 if (!cfg
.test_regex
.empty()){
4169 PrintOut(LOG_INFO
, "File %s line %d (drive %s): ignoring previous Test Directive -s %s\n",
4170 configfile
, lineno
, name
, cfg
.test_regex
.get_pattern());
4171 cfg
.test_regex
= regular_expression();
4173 // check for missing argument
4174 if (!(arg
= strtok(NULL
, delim
))) {
4179 if (!cfg
.test_regex
.compile(arg
, REG_EXTENDED
)) {
4180 // not a valid regular expression!
4181 PrintOut(LOG_CRIT
, "File %s line %d (drive %s): -s argument \"%s\" is INVALID extended regular expression. %s.\n",
4182 configfile
, lineno
, name
, arg
, cfg
.test_regex
.get_errmsg());
4185 // Do a bit of sanity checking and warn user if we think that
4186 // their regexp is "strange". User probably confused about shell
4187 // glob(3) syntax versus regular expression syntax regexp(7).
4188 if (arg
[(val
= strspn(arg
, "0123456789/.-+*|()?^$[]SLCOcnr"))])
4189 PrintOut(LOG_INFO
, "File %s line %d (drive %s): warning, character %d (%c) looks odd in extended regular expression %s\n",
4190 configfile
, lineno
, name
, val
+1, arg
[val
], arg
);
4194 // send email to address that follows
4195 if (!(arg
= strtok(NULL
,delim
)))
4198 if (!cfg
.emailaddress
.empty())
4199 PrintOut(LOG_INFO
, "File %s line %d (drive %s): ignoring previous Address Directive -m %s\n",
4200 configfile
, lineno
, name
, cfg
.emailaddress
.c_str());
4201 #ifdef _WIN32 // TODO: Remove after smartmontools 6.5
4202 if ( !strcmp(arg
, "msgbox") || !strcmp(arg
, "sysmsgbox")
4203 || str_starts_with(arg
, "msgbox,") || str_starts_with(arg
, "sysmsgbox,")) {
4204 PrintOut(LOG_CRIT
, "File %s line %d (drive %s): -m %s is no longer supported, use -m console[,...] instead\n",
4205 configfile
, lineno
, name
, arg
);
4209 cfg
.emailaddress
= arg
;
4213 // email warning options
4214 if (!(arg
= strtok(NULL
, delim
)))
4216 else if (!strcmp(arg
, "once"))
4218 else if (!strcmp(arg
, "daily"))
4220 else if (!strcmp(arg
, "diminishing"))
4222 else if (!strcmp(arg
, "test"))
4224 else if (!strcmp(arg
, "exec")) {
4225 // Get the next argument (the command line)
4227 // Allow "/path name/with spaces/..." on Windows
4228 arg
= strtok_dequote(delim
);
4229 if (arg
&& arg
[0] == '"') {
4230 PrintOut(LOG_CRIT
, "File %s line %d (drive %s): Directive %s 'exec' argument: missing closing quote\n",
4231 configfile
, lineno
, name
, token
);
4235 arg
= strtok(0, delim
);
4238 PrintOut(LOG_CRIT
, "File %s line %d (drive %s): Directive %s 'exec' argument must be followed by executable path.\n",
4239 configfile
, lineno
, name
, token
);
4242 // Free the last cmd line given if any, and copy new one
4243 if (!cfg
.emailcmdline
.empty())
4244 PrintOut(LOG_INFO
, "File %s line %d (drive %s): ignoring previous mail Directive -M exec %s\n",
4245 configfile
, lineno
, name
, cfg
.emailcmdline
.c_str());
4246 cfg
.emailcmdline
= arg
;
4252 // ignore failure of usage attribute
4253 if ((val
=GetInteger(arg
=strtok(NULL
,delim
), name
, token
, lineno
, configfile
, 1, 255))<0)
4255 cfg
.monitor_attr_flags
.set(val
, MONITOR_IGN_FAILUSE
);
4258 // ignore attribute for tracking purposes
4259 if ((val
=GetInteger(arg
=strtok(NULL
,delim
), name
, token
, lineno
, configfile
, 1, 255))<0)
4261 cfg
.monitor_attr_flags
.set(val
, MONITOR_IGNORE
);
4264 // print raw value when tracking
4265 if ((val
= GetInteger(arg
=strtok(NULL
,delim
), name
, token
, lineno
, configfile
, 1, 255, excl
)) < 0)
4267 cfg
.monitor_attr_flags
.set(val
, MONITOR_RAW_PRINT
);
4268 if (*excl
== '!') // attribute change is critical
4269 cfg
.monitor_attr_flags
.set(val
, MONITOR_AS_CRIT
);
4272 // track changes in raw value (forces printing of raw value)
4273 if ((val
= GetInteger(arg
=strtok(NULL
,delim
), name
, token
, lineno
, configfile
, 1, 255, excl
)) < 0)
4275 cfg
.monitor_attr_flags
.set(val
, MONITOR_RAW_PRINT
|MONITOR_RAW
);
4276 if (*excl
== '!') // raw value change is critical
4277 cfg
.monitor_attr_flags
.set(val
, MONITOR_RAW_AS_CRIT
);
4280 // track Temperature
4281 if (Get3Integers(arg
=strtok(NULL
, delim
), name
, token
, lineno
, configfile
,
4282 &cfg
.tempdiff
, &cfg
.tempinfo
, &cfg
.tempcrit
) < 0)
4286 // non-default vendor-specific attribute meaning
4287 if (!(arg
=strtok(NULL
,delim
))) {
4289 } else if (!parse_attribute_def(arg
, cfg
.attribute_defs
, PRIOR_USER
)) {
4294 // Define use of drive-specific presets.
4295 if (!(arg
= strtok(NULL
, delim
))) {
4297 } else if (!strcmp(arg
, "use")) {
4298 cfg
.ignorepresets
= false;
4299 } else if (!strcmp(arg
, "ignore")) {
4300 cfg
.ignorepresets
= true;
4301 } else if (!strcmp(arg
, "show")) {
4302 cfg
.showpresets
= true;
4303 } else if (!strcmp(arg
, "showall")) {
4311 // Various ATA settings
4312 if (!(arg
= strtok(NULL
, delim
))) {
4316 char arg2
[16+1]; unsigned val
;
4317 int n1
= -1, n2
= -1, n3
= -1, len
= strlen(arg
);
4318 if (sscanf(arg
, "%16[^,=]%n%*[,=]%n%u%n", arg2
, &n1
, &n2
, &val
, &n3
) >= 1
4319 && (n1
== len
|| n2
> 0)) {
4320 bool on
= (n2
> 0 && !strcmp(arg
+n2
, "on"));
4321 bool off
= (n2
> 0 && !strcmp(arg
+n2
, "off"));
4325 if (!strcmp(arg2
, "aam")) {
4328 else if (val
<= 254)
4329 cfg
.set_aam
= val
+ 1;
4333 else if (!strcmp(arg2
, "apm")) {
4336 else if (1 <= val
&& val
<= 254)
4337 cfg
.set_apm
= val
+ 1;
4341 else if (!strcmp(arg2
, "lookahead")) {
4343 cfg
.set_lookahead
= -1;
4345 cfg
.set_lookahead
= 1;
4349 else if (!strcmp(arg
, "security-freeze")) {
4350 cfg
.set_security_freeze
= true;
4352 else if (!strcmp(arg2
, "standby")) {
4354 cfg
.set_standby
= 0 + 1;
4355 else if (val
<= 255)
4356 cfg
.set_standby
= val
+ 1;
4360 else if (!strcmp(arg2
, "wcache")) {
4362 cfg
.set_wcache
= -1;
4377 // Directive not recognized
4378 PrintOut(LOG_CRIT
,"File %s line %d (drive %s): unknown Directive: %s\n",
4379 configfile
, lineno
, name
, token
);
4384 PrintOut(LOG_CRIT
, "File %s line %d (drive %s): Missing argument to %s Directive\n",
4385 configfile
, lineno
, name
, token
);
4388 PrintOut(LOG_CRIT
, "File %s line %d (drive %s): Invalid argument to %s Directive: %s\n",
4389 configfile
, lineno
, name
, token
, arg
);
4391 if (missingarg
|| badarg
) {
4392 PrintOut(LOG_CRIT
, "Valid arguments to %s Directive are: ", token
);
4393 printoutvaliddirectiveargs(LOG_CRIT
, sym
);
4394 PrintOut(LOG_CRIT
, "\n");
4401 // Scan directive for configuration file
4402 #define SCANDIRECTIVE "DEVICESCAN"
4404 // This is the routine that adds things to the conf_entries list.
4406 // Return values are:
4407 // 1: parsed a normal line
4408 // 0: found DEFAULT setting or comment or blank line
4409 // -1: found SCANDIRECTIVE line
4410 // -2: found an error
4412 // Note: this routine modifies *line from the caller!
4413 static int ParseConfigLine(dev_config_vector
& conf_entries
, dev_config
& default_conf
,
4414 smart_devtype_list
& scan_types
, int lineno
, /*const*/ char * line
)
4416 const char *delim
= " \n\t";
4418 // get first token: device name. If a comment, skip line
4419 const char * name
= strtok(line
, delim
);
4420 if (!name
|| *name
== '#')
4423 // Check device name for DEFAULT or DEVICESCAN
4425 if (!strcmp("DEFAULT", name
)) {
4427 // Restart with empty defaults
4428 default_conf
= dev_config();
4431 retval
= (!strcmp(SCANDIRECTIVE
, name
) ? -1 : 1);
4432 // Init new entry with current defaults
4433 conf_entries
.push_back(default_conf
);
4435 dev_config
& cfg
= (retval
? conf_entries
.back() : default_conf
);
4437 cfg
.name
= name
; // Later replaced by dev->get_info().info_name
4438 cfg
.dev_name
= name
; // If DEVICESCAN later replaced by get->dev_info().dev_name
4439 cfg
.lineno
= lineno
;
4441 // parse tokens one at a time from the file.
4442 while (char * token
= strtok(0, delim
)) {
4443 int rc
= ParseToken(token
, cfg
, scan_types
);
4445 // error found on the line
4452 // PrintOut(LOG_INFO,"Parsed token %s\n",token);
4455 // Check for multiple -d TYPE directives
4456 if (retval
!= -1 && scan_types
.size() > 1) {
4457 PrintOut(LOG_CRIT
, "Drive: %s, invalid multiple -d TYPE Directives on line %d of file %s\n",
4458 cfg
.name
.c_str(), cfg
.lineno
, configfile
);
4462 // Don't perform checks below for DEFAULT entries
4466 // If NO monitoring directives are set, then set all of them.
4467 if (!( cfg
.smartcheck
|| cfg
.selftest
4468 || cfg
.errorlog
|| cfg
.xerrorlog
4469 || cfg
.offlinests
|| cfg
.selfteststs
4470 || cfg
.usagefailed
|| cfg
.prefail
|| cfg
.usage
4471 || cfg
.tempdiff
|| cfg
.tempinfo
|| cfg
.tempcrit
)) {
4473 PrintOut(LOG_INFO
,"Drive: %s, implied '-a' Directive on line %d of file %s\n",
4474 cfg
.name
.c_str(), cfg
.lineno
, configfile
);
4476 cfg
.smartcheck
= true;
4477 cfg
.usagefailed
= true;
4480 cfg
.selftest
= true;
4481 cfg
.errorlog
= true;
4482 cfg
.selfteststs
= true;
4485 // additional sanity check. Has user set -M options without -m?
4486 if (cfg
.emailaddress
.empty() && (!cfg
.emailcmdline
.empty() || cfg
.emailfreq
|| cfg
.emailtest
)){
4487 PrintOut(LOG_CRIT
,"Drive: %s, -M Directive(s) on line %d of file %s need -m ADDRESS Directive\n",
4488 cfg
.name
.c_str(), cfg
.lineno
, configfile
);
4492 // has the user has set <nomailer>?
4493 if (cfg
.emailaddress
== "<nomailer>") {
4494 // check that -M exec is also set
4495 if (cfg
.emailcmdline
.empty()){
4496 PrintOut(LOG_CRIT
,"Drive: %s, -m <nomailer> Directive on line %d of file %s needs -M exec Directive\n",
4497 cfg
.name
.c_str(), cfg
.lineno
, configfile
);
4500 // From here on the sign of <nomailer> is cfg.emailaddress.empty() and !cfg.emailcmdline.empty()
4501 cfg
.emailaddress
.clear();
4507 // Parses a configuration file. Return values are:
4508 // N=>0: found N entries
4509 // -1: syntax error in config file
4510 // -2: config file does not exist
4511 // -3: config file exists but cannot be read
4513 // In the case where the return value is 0, there are three
4515 // Empty configuration file ==> conf_entries.empty()
4516 // No configuration file ==> conf_entries[0].lineno == 0
4517 // SCANDIRECTIVE found ==> conf_entries.back().lineno != 0 (size >= 1)
4518 static int ParseConfigFile(dev_config_vector
& conf_entries
, smart_devtype_list
& scan_types
)
4520 // maximum line length in configuration file
4521 const int MAXLINELEN
= 256;
4522 // maximum length of a continued line in configuration file
4523 const int MAXCONTLINE
= 1023;
4526 // Open config file, if it exists and is not <stdin>
4527 if (!(configfile
== configfile_stdin
)) { // pointer comparison ok here
4528 if (!f
.open(configfile
,"r") && (errno
!=ENOENT
|| !configfile_alt
.empty())) {
4529 // file exists but we can't read it or it should exist due to '-c' option
4530 int ret
= (errno
!=ENOENT
? -3 : -2);
4531 PrintOut(LOG_CRIT
,"%s: Unable to open configuration file %s\n",
4532 strerror(errno
),configfile
);
4536 else // read from stdin ('-c -' option)
4539 // Start with empty defaults
4540 dev_config default_conf
;
4542 // No configuration file found -- use fake one
4545 char fakeconfig
[] = SCANDIRECTIVE
" -a"; // TODO: Remove this hack, build cfg_entry.
4547 if (ParseConfigLine(conf_entries
, default_conf
, scan_types
, 0, fakeconfig
) != -1)
4548 throw std::logic_error("Internal error parsing " SCANDIRECTIVE
);
4553 setmode(fileno(f
), O_TEXT
); // Allow files with \r\n
4556 // configuration file exists
4557 PrintOut(LOG_INFO
,"Opened configuration file %s\n",configfile
);
4559 // parse config file line by line
4560 int lineno
= 1, cont
= 0, contlineno
= 0;
4561 char line
[MAXLINELEN
+2];
4562 char fullline
[MAXCONTLINE
+1];
4565 int len
=0,scandevice
;
4570 // make debugging simpler
4571 memset(line
,0,sizeof(line
));
4574 code
=fgets(line
, MAXLINELEN
+2, f
);
4576 // are we at the end of the file?
4579 scandevice
= ParseConfigLine(conf_entries
, default_conf
, scan_types
, contlineno
, fullline
);
4580 // See if we found a SCANDIRECTIVE directive
4583 // did we find a syntax error
4586 // the final line is part of a continuation line
4592 // input file line number
4595 // See if line is too long
4597 if (len
>MAXLINELEN
){
4599 if (line
[len
-1]=='\n')
4600 warn
="(including newline!) ";
4603 PrintOut(LOG_CRIT
,"Error: line %d of file %s %sis more than MAXLINELEN=%d characters.\n",
4604 (int)contlineno
,configfile
,warn
,(int)MAXLINELEN
);
4608 // Ignore anything after comment symbol
4609 if ((comment
=strchr(line
,'#'))){
4614 // is the total line (made of all continuation lines) too long?
4615 if (cont
+len
>MAXCONTLINE
){
4616 PrintOut(LOG_CRIT
,"Error: continued line %d (actual line %d) of file %s is more than MAXCONTLINE=%d characters.\n",
4617 lineno
, (int)contlineno
, configfile
, (int)MAXCONTLINE
);
4621 // copy string so far into fullline, and increment length
4622 snprintf(fullline
+cont
, sizeof(fullline
)-cont
, "%s" ,line
);
4625 // is this a continuation line. If so, replace \ by space and look at next line
4626 if ( (lastslash
=strrchr(line
,'\\')) && !strtok(lastslash
+1," \n\t")){
4627 *(fullline
+(cont
-len
)+(lastslash
-line
))=' ';
4631 // Not a continuation line. Parse it
4633 scandevice
= ParseConfigLine(conf_entries
, default_conf
, scan_types
, contlineno
, fullline
);
4635 // did we find a scandevice directive?
4638 // did we find a syntax error
4647 // note -- may be zero if syntax of file OK, but no valid entries!
4651 /* Prints the message "=======> VALID ARGUMENTS ARE: <LIST> <=======\n", where
4652 <LIST> is the list of valid arguments for option opt. */
4653 static void PrintValidArgs(char opt
)
4657 PrintOut(LOG_CRIT
, "=======> VALID ARGUMENTS ARE: ");
4658 if (!(s
= GetValidArgList(opt
)))
4659 PrintOut(LOG_CRIT
, "Error constructing argument list for option %c", opt
);
4661 PrintOut(LOG_CRIT
, "%s", (char *)s
);
4662 PrintOut(LOG_CRIT
, " <=======\n");
4666 // Report error and exit if specified path is not absolute.
4667 static void check_abs_path(char option
, const std::string
& path
)
4669 if (path
.empty() || path
[0] == '/')
4674 PrintOut(LOG_CRIT
, "=======> INVALID ARGUMENT TO -%c: %s <=======\n\n", option
, path
.c_str());
4675 PrintOut(LOG_CRIT
, "Error: relative path names are not allowed\n\n");
4680 // Parses input line, prints usage message and
4681 // version/license/copyright messages
4682 static void ParseOpts(int argc
, char **argv
)
4684 // Init default path names
4686 configfile
= SMARTMONTOOLS_SYSCONFDIR
"/smartd.conf";
4687 warning_script
= SMARTMONTOOLS_SMARTDSCRIPTDIR
"/smartd_warning.sh";
4689 std::string exedir
= get_exe_dir();
4690 static std::string configfile_str
= exedir
+ "/smartd.conf";
4691 configfile
= configfile_str
.c_str();
4692 warning_script
= exedir
+ "/smartd_warning.cmd";
4695 // Please update GetValidArgList() if you edit shortopts
4696 static const char shortopts
[] = "c:l:q:dDni:p:r:s:A:B:w:Vh?"
4697 #ifdef HAVE_LIBCAP_NG
4701 // Please update GetValidArgList() if you edit longopts
4702 struct option longopts
[] = {
4703 { "configfile", required_argument
, 0, 'c' },
4704 { "logfacility", required_argument
, 0, 'l' },
4705 { "quit", required_argument
, 0, 'q' },
4706 { "debug", no_argument
, 0, 'd' },
4707 { "showdirectives", no_argument
, 0, 'D' },
4708 { "interval", required_argument
, 0, 'i' },
4710 { "no-fork", no_argument
, 0, 'n' },
4712 { "service", no_argument
, 0, 'n' },
4714 { "pidfile", required_argument
, 0, 'p' },
4715 { "report", required_argument
, 0, 'r' },
4716 { "savestates", required_argument
, 0, 's' },
4717 { "attributelog", required_argument
, 0, 'A' },
4718 { "drivedb", required_argument
, 0, 'B' },
4719 { "warnexec", required_argument
, 0, 'w' },
4720 { "version", no_argument
, 0, 'V' },
4721 { "license", no_argument
, 0, 'V' },
4722 { "copyright", no_argument
, 0, 'V' },
4723 { "help", no_argument
, 0, 'h' },
4724 { "usage", no_argument
, 0, 'h' },
4725 #ifdef HAVE_LIBCAP_NG
4726 { "capabilities", no_argument
, 0, 'C' },
4732 bool badarg
= false;
4733 bool use_default_db
= true; // set false on '-B FILE'
4735 // Parse input options.
4737 while ((optchar
= getopt_long(argc
, argv
, shortopts
, longopts
, NULL
)) != -1) {
4745 if (!(strcmp(optarg
,"nodev"))) {
4747 } else if (!(strcmp(optarg
,"nodevstartup"))) {
4749 } else if (!(strcmp(optarg
,"never"))) {
4751 } else if (!(strcmp(optarg
,"onecheck"))) {
4754 } else if (!(strcmp(optarg
,"showtests"))) {
4757 } else if (!(strcmp(optarg
,"errors"))) {
4764 // set the log facility level
4765 if (!strcmp(optarg
, "daemon"))
4766 facility
=LOG_DAEMON
;
4767 else if (!strcmp(optarg
, "local0"))
4768 facility
=LOG_LOCAL0
;
4769 else if (!strcmp(optarg
, "local1"))
4770 facility
=LOG_LOCAL1
;
4771 else if (!strcmp(optarg
, "local2"))
4772 facility
=LOG_LOCAL2
;
4773 else if (!strcmp(optarg
, "local3"))
4774 facility
=LOG_LOCAL3
;
4775 else if (!strcmp(optarg
, "local4"))
4776 facility
=LOG_LOCAL4
;
4777 else if (!strcmp(optarg
, "local5"))
4778 facility
=LOG_LOCAL5
;
4779 else if (!strcmp(optarg
, "local6"))
4780 facility
=LOG_LOCAL6
;
4781 else if (!strcmp(optarg
, "local7"))
4782 facility
=LOG_LOCAL7
;
4787 // enable debug mode
4792 #ifndef _WIN32 // On Windows, --service is already handled by daemon_main()
4797 // print summary of all valid directives
4803 // Period (time interval) for checking
4804 // strtol will set errno in the event of overflow, so we'll check it.
4806 lchecktime
= strtol(optarg
, &tailptr
, 10);
4807 if (*tailptr
!= '\0' || lchecktime
< 10 || lchecktime
> INT_MAX
|| errno
) {
4810 PrintOut(LOG_CRIT
, "======> INVALID INTERVAL: %s <=======\n", optarg
);
4811 PrintOut(LOG_CRIT
, "======> INTERVAL MUST BE INTEGER BETWEEN %d AND %d <=======\n", 10, INT_MAX
);
4812 PrintOut(LOG_CRIT
, "\nUse smartd -h to get a usage summary\n\n");
4815 checktime
= (int)lchecktime
;
4818 // report IOCTL transactions
4820 int n1
= -1, n2
= -1, len
= strlen(optarg
);
4821 char s
[9+1]; unsigned i
= 1;
4822 sscanf(optarg
, "%9[a-z]%n,%u%n", s
, &n1
, &i
, &n2
);
4823 if (!((n1
== len
|| n2
== len
) && 1 <= i
&& i
<= 4)) {
4825 } else if (!strcmp(s
,"ioctl")) {
4826 ata_debugmode
= scsi_debugmode
= nvme_debugmode
= i
;
4827 } else if (!strcmp(s
,"ataioctl")) {
4829 } else if (!strcmp(s
,"scsiioctl")) {
4831 } else if (!strcmp(s
,"nvmeioctl")) {
4839 // alternate configuration file
4840 if (strcmp(optarg
,"-"))
4841 configfile
= (configfile_alt
= optarg
).c_str();
4842 else // read from stdin
4843 configfile
=configfile_stdin
;
4846 // output file with PID number
4850 // path prefix of persistent state file
4851 state_path_prefix
= optarg
;
4854 // path prefix of attribute log file
4855 attrlog_path_prefix
= optarg
;
4859 const char * path
= optarg
;
4860 if (*path
== '+' && path
[1])
4863 use_default_db
= false;
4864 unsigned char savedebug
= debugmode
; debugmode
= 1;
4865 if (!read_drive_database(path
))
4867 debugmode
= savedebug
;
4871 warning_script
= optarg
;
4874 // print version and CVS info
4876 PrintOut(LOG_INFO
, "%s", format_version_info("smartd", true /*full*/).c_str());
4879 #ifdef HAVE_LIBCAP_NG
4881 // enable capabilities
4882 enable_capabilities
= true;
4886 // help: print summary of command-line options
4894 // unrecognized option
4897 // Point arg to the argument in which this option was found.
4898 arg
= argv
[optind
-1];
4899 // Check whether the option is a long option that doesn't map to -h.
4900 if (arg
[1] == '-' && optchar
!= 'h') {
4901 // Iff optopt holds a valid option then argument must be missing.
4902 if (optopt
&& (strchr(shortopts
, optopt
) != NULL
)) {
4903 PrintOut(LOG_CRIT
, "=======> ARGUMENT REQUIRED FOR OPTION: %s <=======\n",arg
+2);
4904 PrintValidArgs(optopt
);
4906 PrintOut(LOG_CRIT
, "=======> UNRECOGNIZED OPTION: %s <=======\n\n",arg
+2);
4908 PrintOut(LOG_CRIT
, "\nUse smartd --help to get a usage summary\n\n");
4912 // Iff optopt holds a valid option then argument must be missing.
4913 if (strchr(shortopts
, optopt
) != NULL
){
4914 PrintOut(LOG_CRIT
, "=======> ARGUMENT REQUIRED FOR OPTION: %c <=======\n",optopt
);
4915 PrintValidArgs(optopt
);
4917 PrintOut(LOG_CRIT
, "=======> UNRECOGNIZED OPTION: %c <=======\n\n",optopt
);
4919 PrintOut(LOG_CRIT
, "\nUse smartd -h to get a usage summary\n\n");
4926 // Check to see if option had an unrecognized or incorrect argument.
4930 // It would be nice to print the actual option name given by the user
4931 // here, but we just print the short form. Please fix this if you know
4932 // a clean way to do it.
4933 PrintOut(LOG_CRIT
, "=======> INVALID ARGUMENT TO -%c: %s <======= \n", optchar
, optarg
);
4934 PrintValidArgs(optchar
);
4935 PrintOut(LOG_CRIT
, "\nUse smartd -h to get a usage summary\n\n");
4940 // non-option arguments are not allowed
4941 if (argc
> optind
) {
4944 PrintOut(LOG_CRIT
, "=======> UNRECOGNIZED ARGUMENT: %s <=======\n\n", argv
[optind
]);
4945 PrintOut(LOG_CRIT
, "\nUse smartd -h to get a usage summary\n\n");
4949 // no pidfile in debug mode
4950 if (debugmode
&& !pid_file
.empty()) {
4953 PrintOut(LOG_CRIT
, "=======> INVALID CHOICE OF OPTIONS: -d and -p <======= \n\n");
4954 PrintOut(LOG_CRIT
, "Error: pid file %s not written in debug (-d) mode\n\n", pid_file
.c_str());
4960 // absolute path names are required due to chdir('/') after fork().
4961 check_abs_path('p', pid_file
);
4962 check_abs_path('s', state_path_prefix
);
4963 check_abs_path('A', attrlog_path_prefix
);
4967 // Read or init drive database
4969 unsigned char savedebug
= debugmode
; debugmode
= 1;
4970 if (!init_drive_database(use_default_db
))
4972 debugmode
= savedebug
;
4979 // Function we call if no configuration file was found or if the
4980 // SCANDIRECTIVE Directive was found. It makes entries for device
4981 // names returned by scan_smart_devices() in os_OSNAME.cpp
4982 static int MakeConfigEntries(const dev_config
& base_cfg
,
4983 dev_config_vector
& conf_entries
, smart_device_list
& scanned_devs
,
4984 const smart_devtype_list
& types
)
4986 // make list of devices
4987 smart_device_list devlist
;
4988 if (!smi()->scan_smart_devices(devlist
, types
)) {
4989 PrintOut(LOG_CRIT
, "DEVICESCAN failed: %s\n", smi()->get_errmsg());
4993 // if no devices, return
4994 if (devlist
.size() <= 0)
4997 // add empty device slots for existing config entries
4998 while (scanned_devs
.size() < conf_entries
.size())
4999 scanned_devs
.push_back((smart_device
*)0);
5001 // loop over entries to create
5002 for (unsigned i
= 0; i
< devlist
.size(); i
++) {
5003 // Move device pointer
5004 smart_device
* dev
= devlist
.release(i
);
5005 scanned_devs
.push_back(dev
);
5007 // Copy configuration, update device and type name
5008 conf_entries
.push_back(base_cfg
);
5009 dev_config
& cfg
= conf_entries
.back();
5010 cfg
.name
= dev
->get_info().info_name
;
5011 cfg
.dev_name
= dev
->get_info().dev_name
;
5012 cfg
.dev_type
= dev
->get_info().dev_type
;
5015 return devlist
.size();
5018 static void CanNotRegister(const char *name
, const char *type
, int line
, bool scandirective
)
5020 if (!debugmode
&& scandirective
)
5023 PrintOut(scandirective
?LOG_INFO
:LOG_CRIT
,
5024 "Unable to register %s device %s at line %d of file %s\n",
5025 type
, name
, line
, configfile
);
5027 PrintOut(LOG_INFO
,"Unable to register %s device %s\n",
5032 // Returns negative value (see ParseConfigFile()) if config file
5033 // had errors, else number of entries which may be zero or positive.
5034 static int ReadOrMakeConfigEntries(dev_config_vector
& conf_entries
, smart_device_list
& scanned_devs
)
5036 // parse configuration file configfile (normally /etc/smartd.conf)
5037 smart_devtype_list scan_types
;
5038 int entries
= ParseConfigFile(conf_entries
, scan_types
);
5041 // There was an error reading the configuration file.
5042 conf_entries
.clear();
5044 PrintOut(LOG_CRIT
, "Configuration file %s has fatal syntax errors.\n", configfile
);
5048 // no error parsing config file.
5050 // we did not find a SCANDIRECTIVE and did find valid entries
5051 PrintOut(LOG_INFO
, "Configuration file %s parsed.\n", configfile
);
5053 else if (!conf_entries
.empty()) {
5054 // we found a SCANDIRECTIVE or there was no configuration file so
5055 // scan. Configuration file's last entry contains all options
5057 dev_config first
= conf_entries
.back();
5058 conf_entries
.pop_back();
5061 PrintOut(LOG_INFO
,"Configuration file %s was parsed, found %s, scanning devices\n", configfile
, SCANDIRECTIVE
);
5063 PrintOut(LOG_INFO
,"No configuration file %s found, scanning devices\n", configfile
);
5065 // make config list of devices to search for
5066 MakeConfigEntries(first
, conf_entries
, scanned_devs
, scan_types
);
5068 // warn user if scan table found no devices
5069 if (conf_entries
.empty())
5070 PrintOut(LOG_CRIT
,"In the system's table of devices NO devices found to scan\n");
5073 PrintOut(LOG_CRIT
, "Configuration file %s parsed but has no entries\n", configfile
);
5075 return conf_entries
.size();
5078 // Return true if TYPE contains a RAID drive number
5079 static bool is_raid_type(const char * type
)
5081 if (str_starts_with(type
, "sat,"))
5084 if (sscanf(type
, "%*[^,],%d", &i
) != 1)
5089 // Return true if DEV is already in DEVICES[0..NUMDEVS) or IGNORED[*]
5090 static bool is_duplicate_device(const smart_device
* dev
,
5091 const smart_device_list
& devices
, unsigned numdevs
,
5092 const dev_config_vector
& ignored
)
5094 const smart_device::device_info
& info1
= dev
->get_info();
5095 bool is_raid1
= is_raid_type(info1
.dev_type
.c_str());
5097 for (unsigned i
= 0; i
< numdevs
; i
++) {
5098 const smart_device::device_info
& info2
= devices
.at(i
)->get_info();
5099 // -d TYPE options must match if RAID drive number is specified
5100 if ( info1
.dev_name
== info2
.dev_name
5101 && ( info1
.dev_type
== info2
.dev_type
5102 || !is_raid1
|| !is_raid_type(info2
.dev_type
.c_str())))
5106 for (unsigned i
= 0; i
< ignored
.size(); i
++) {
5107 const dev_config
& cfg2
= ignored
.at(i
);
5108 if ( info1
.dev_name
== cfg2
.dev_name
5109 && ( info1
.dev_type
== cfg2
.dev_type
5110 || !is_raid1
|| !is_raid_type(cfg2
.dev_type
.c_str())))
5116 // This function tries devices from conf_entries. Each one that can be
5117 // registered is moved onto the [ata|scsi]devices lists and removed
5118 // from the conf_entries list.
5119 static void RegisterDevices(const dev_config_vector
& conf_entries
, smart_device_list
& scanned_devs
,
5120 dev_config_vector
& configs
, dev_state_vector
& states
, smart_device_list
& devices
)
5122 // start by clearing lists/memory of ALL existing devices
5128 dev_config_vector ignored_entries
;
5129 unsigned numnoscan
= 0;
5130 for (unsigned i
= 0; i
< conf_entries
.size(); i
++){
5132 dev_config cfg
= conf_entries
[i
];
5135 // Store for is_duplicate_device() check and ignore
5136 PrintOut(LOG_INFO
, "Device: %s%s%s%s, ignored\n", cfg
.name
.c_str(),
5137 (!cfg
.dev_type
.empty() ? " [" : ""),
5138 cfg
.dev_type
.c_str(),
5139 (!cfg
.dev_type
.empty() ? "]" : ""));
5140 ignored_entries
.push_back(cfg
);
5144 // get device of appropriate type
5145 smart_device_auto_ptr dev
;
5146 bool scanning
= false;
5148 // Device may already be detected during devicescan
5149 if (i
< scanned_devs
.size()) {
5150 dev
= scanned_devs
.release(i
);
5152 // Check for a preceding non-DEVICESCAN entry for the same device
5153 if ( (numnoscan
|| !ignored_entries
.empty())
5154 && is_duplicate_device(dev
.get(), devices
, numnoscan
, ignored_entries
)) {
5155 PrintOut(LOG_INFO
, "Device: %s, duplicate, ignored\n", dev
->get_info_name());
5163 dev
= smi()->get_smart_device(cfg
.name
.c_str(), cfg
.dev_type
.c_str());
5165 if (cfg
.dev_type
.empty())
5166 PrintOut(LOG_INFO
,"Device: %s, unable to autodetect device type\n", cfg
.name
.c_str());
5168 PrintOut(LOG_INFO
,"Device: %s, unsupported device type '%s'\n", cfg
.name
.c_str(), cfg
.dev_type
.c_str());
5174 smart_device::device_info oldinfo
= dev
->get_info();
5176 // Open with autodetect support, may return 'better' device
5177 dev
.replace( dev
->autodetect_open() );
5179 // Report if type has changed
5180 if (oldinfo
.dev_type
!= dev
->get_dev_type())
5181 PrintOut(LOG_INFO
,"Device: %s, type changed from '%s' to '%s'\n",
5182 cfg
.name
.c_str(), oldinfo
.dev_type
.c_str(), dev
->get_dev_type());
5184 if (!dev
->is_open()) {
5185 // For linux+devfs, a nonexistent device gives a strange error
5186 // message. This makes the error message a bit more sensible.
5187 // If no debug and scanning - don't print errors
5188 if (debugmode
|| !scanning
)
5189 PrintOut(LOG_INFO
, "Device: %s, open() failed: %s\n", dev
->get_info_name(), dev
->get_errmsg());
5193 // Update informal name
5194 cfg
.name
= dev
->get_info().info_name
;
5195 PrintOut(LOG_INFO
, "Device: %s, opened\n", cfg
.name
.c_str());
5197 // Prepare initial state
5200 // register ATA devices
5202 if (ATADeviceScan(cfg
, state
, dev
->to_ata())) {
5203 CanNotRegister(cfg
.name
.c_str(), "ATA", cfg
.lineno
, scanning
);
5207 // or register SCSI devices
5208 else if (dev
->is_scsi()){
5209 if (SCSIDeviceScan(cfg
, state
, dev
->to_scsi())) {
5210 CanNotRegister(cfg
.name
.c_str(), "SCSI", cfg
.lineno
, scanning
);
5214 // or register NVMe devices
5215 else if (dev
->is_nvme()) {
5216 if (NVMeDeviceScan(cfg
, state
, dev
->to_nvme())) {
5217 CanNotRegister(cfg
.name
.c_str(), "NVMe", cfg
.lineno
, scanning
);
5222 PrintOut(LOG_INFO
, "Device: %s, neither ATA, SCSI nor NVMe device\n", cfg
.name
.c_str());
5227 // move onto the list of devices
5228 configs
.push_back(cfg
);
5229 states
.push_back(state
);
5230 devices
.push_back(dev
);
5232 numnoscan
= devices
.size();
5234 // if device is explictly listed and we can't register it, then
5235 // exit unless the user has specified that the device is removable
5236 else if (!scanning
) {
5237 if (cfg
.removable
|| quit
==2)
5238 PrintOut(LOG_INFO
, "Device %s not available\n", cfg
.name
.c_str());
5240 PrintOut(LOG_CRIT
, "Unable to register device %s (no Directive -d removable). Exiting.\n", cfg
.name
.c_str());
5246 init_disable_standby_check(configs
);
5250 // Main program without exception handling
5251 static int main_worker(int argc
, char **argv
)
5253 // Initialize interface
5254 smart_interface::init();
5258 // is it our first pass through?
5259 bool firstpass
= true;
5261 // next time to wake up
5262 time_t wakeuptime
= 0;
5264 // parse input and print header and usage info if needed
5265 ParseOpts(argc
,argv
);
5267 // Configuration for each device
5268 dev_config_vector configs
;
5270 dev_state_vector states
;
5271 // Devices to monitor
5272 smart_device_list devices
;
5274 bool write_states_always
= true;
5276 #ifdef HAVE_LIBCAP_NG
5277 // Drop capabilities
5278 if (enable_capabilities
) {
5279 capng_clear(CAPNG_SELECT_BOTH
);
5280 capng_updatev(CAPNG_ADD
, (capng_type_t
)(CAPNG_EFFECTIVE
|CAPNG_PERMITTED
),
5281 CAP_SYS_ADMIN
, CAP_MKNOD
, CAP_SYS_RAWIO
, -1);
5282 capng_apply(CAPNG_SELECT_BOTH
);
5286 // the main loop of the code
5289 // are we exiting from a signal?
5290 if (caughtsigEXIT
) {
5291 // are we exiting with SIGTERM?
5292 int isterm
=(caughtsigEXIT
==SIGTERM
);
5293 int isquit
=(caughtsigEXIT
==SIGQUIT
);
5294 int isok
=debugmode
?isterm
|| isquit
:isterm
;
5296 PrintOut(isok
?LOG_INFO
:LOG_CRIT
, "smartd received signal %d: %s\n",
5297 caughtsigEXIT
, strsignal(caughtsigEXIT
));
5302 // Write state files
5303 if (!state_path_prefix
.empty())
5304 write_all_dev_states(configs
, states
);
5309 // Should we (re)read the config file?
5310 if (firstpass
|| caughtsigHUP
){
5312 // Write state files
5313 if (!state_path_prefix
.empty())
5314 write_all_dev_states(configs
, states
);
5318 "Signal HUP - rereading configuration file %s\n":
5319 "\a\nSignal INT - rereading configuration file %s (" SIGQUIT_KEYNAME
" quits)\n\n",
5324 dev_config_vector conf_entries
; // Entries read from smartd.conf
5325 smart_device_list scanned_devs
; // Devices found during scan
5326 // (re)reads config file, makes >=0 entries
5327 int entries
= ReadOrMakeConfigEntries(conf_entries
, scanned_devs
);
5330 // checks devices, then moves onto ata/scsi list or deallocates.
5331 RegisterDevices(conf_entries
, scanned_devs
, configs
, states
, devices
);
5332 if (!(configs
.size() == devices
.size() && configs
.size() == states
.size()))
5333 throw std::logic_error("Invalid result from RegisterDevices");
5335 else if (quit
==2 || ((quit
==0 || quit
==1) && !firstpass
)) {
5336 // user has asked to continue on error in configuration file
5338 PrintOut(LOG_INFO
,"Reusing previous configuration\n");
5341 // exit with configuration file error status
5342 return (entries
==-3 ? EXIT_READCONF
: entries
==-2 ? EXIT_NOCONF
: EXIT_BADCONF
);
5346 // Log number of devices we are monitoring...
5347 if (devices
.size() > 0 || quit
==2 || (quit
==1 && !firstpass
)) {
5348 int numata
= 0, numscsi
= 0;
5349 for (unsigned i
= 0; i
< devices
.size(); i
++) {
5350 const smart_device
* dev
= devices
.at(i
);
5353 else if (dev
->is_scsi())
5356 PrintOut(LOG_INFO
,"Monitoring %d ATA/SATA, %d SCSI/SAS and %d NVMe devices\n",
5357 numata
, numscsi
, (int)devices
.size() - numata
- numscsi
);
5360 PrintOut(LOG_INFO
,"Unable to monitor any SMART enabled devices. Try debug (-d) option. Exiting...\n");
5365 // user has asked to print test schedule
5366 PrintTestSchedule(configs
, states
, devices
);
5370 #ifdef HAVE_LIBCAP_NG
5371 if (enable_capabilities
) {
5372 for (unsigned i
= 0; i
< configs
.size(); i
++) {
5373 if (!configs
[i
].emailaddress
.empty() || !configs
[i
].emailcmdline
.empty()) {
5374 PrintOut(LOG_WARNING
, "Mail can't be enabled together with --capabilities. All mail will be suppressed.\n");
5384 // Always write state files after (re)configuration
5385 write_states_always
= true;
5388 // check all devices once,
5389 // self tests are not started in first pass unless '-q onecheck' is specified
5390 CheckDevicesOnce(configs
, states
, devices
, firstpass
, (!firstpass
|| quit
==3));
5392 // Write state files
5393 if (!state_path_prefix
.empty())
5394 write_all_dev_states(configs
, states
, write_states_always
);
5395 write_states_always
= false;
5397 // Write attribute logs
5398 if (!attrlog_path_prefix
.empty())
5399 write_all_dev_attrlogs(configs
, states
);
5401 // user has asked us to exit after first check
5403 PrintOut(LOG_INFO
,"Started with '-q onecheck' option. All devices sucessfully checked once.\n"
5404 "smartd is exiting (exit status 0)\n");
5408 // fork into background if needed
5409 if (firstpass
&& !debugmode
) {
5413 // set exit and signal handlers, write PID file, set wake-up time
5415 Initialize(&wakeuptime
);
5419 // sleep until next check time, or a signal arrives
5420 wakeuptime
= dosleep(wakeuptime
, write_states_always
);
5427 int main(int argc
, char **argv
)
5429 // Windows: internal main function started direct or by service control manager
5430 static int smartd_main(int argc
, char **argv
)
5435 // Do the real work ...
5436 status
= main_worker(argc
, argv
);
5439 // EXIT(status) arrives here
5442 catch (const std::bad_alloc
& /*ex*/) {
5443 // Memory allocation failed (also thrown by std::operator new)
5444 PrintOut(LOG_CRIT
, "Smartd: Out of memory\n");
5445 status
= EXIT_NOMEM
;
5447 catch (const std::exception
& ex
) {
5448 // Other fatal errors
5449 PrintOut(LOG_CRIT
, "Smartd: Exception: %s\n", ex
.what());
5450 status
= EXIT_BADCODE
;
5453 // Check for remaining device objects
5454 if (smart_device::get_num_objects() != 0) {
5455 PrintOut(LOG_CRIT
, "Smartd: Internal Error: %d device object(s) left at exit.\n",
5456 smart_device::get_num_objects());
5457 status
= EXIT_BADCODE
;
5460 if (status
== EXIT_BADCODE
)
5461 PrintOut(LOG_CRIT
, "Please inform " PACKAGE_BUGREPORT
", including output of smartd -V.\n");
5464 status
= Goodbye(status
);
5467 daemon_winsvc_exitcode
= status
;
5474 // Main function for Windows
5475 int main(int argc
, char **argv
){
5476 // Options for smartd windows service
5477 static const daemon_winsvc_options svc_opts
= {
5478 "--service", // cmd_opt
5479 "smartd", "SmartD Service", // servicename, displayname
5481 "Controls and monitors storage devices using the Self-Monitoring, "
5482 "Analysis and Reporting Technology System (SMART) built into "
5483 "ATA/SATA and SCSI/SAS hard drives and solid-state drives. "
5484 "www.smartmontools.org"
5486 // daemon_main() handles daemon and service specific commands
5487 // and starts smartd_main() direct, from a new process,
5488 // or via service control manager
5489 return daemon_main("smartd", &svc_opts
, smartd_main
, argc
, argv
);