2 * Home page of code is: http://smartmontools.sourceforge.net
4 * Copyright (C) 2002-9 Bruce Allen <smartmontools-support@lists.sourceforge.net>
5 * Copyright (C) 2000 Michael Cornwell <cornwell@acm.org>
6 * Copyright (C) 2008 Oliver Bock <brevilo@users.sourceforge.net>
7 * Copyright (C) 2008-9 Christian Franke <smartmontools-support@lists.sourceforge.net>
9 * This program is free software; you can redistribute it and/or modify
10 * it under the terms of the GNU General Public License as published by
11 * the Free Software Foundation; either version 2, or (at your option)
14 * You should have received a copy of the GNU General Public License
15 * (for example COPYING); If not, see <http://www.gnu.org/licenses/>.
17 * This code was originally developed as a Senior Thesis by Michael Cornwell
18 * at the Concurrent Systems Laboratory (now part of the Storage Systems
19 * Research Center), Jack Baskin School of Engineering, University of
20 * California, Santa Cruz. http://ssrc.soe.ucsc.edu/
25 // TODO: Why is this define necessary?
29 // unconditionally included files
31 #include <sys/types.h>
32 #include <sys/stat.h> // umask
47 #include <algorithm> // std::replace()
49 // see which system files to conditionally include
52 // conditionally included files
65 #pragma warning(disable:4761) // "conversion supplied"
66 typedef unsigned short mode_t
;
69 #include <io.h> // umask()
70 #include <process.h> // getpid()
75 // BOOL WINAPI FreeConsole(void);
76 extern "C" int __stdcall
FreeConsole(void);
77 #include <io.h> // setmode()
80 // locally included files
83 #include "dev_interface.h"
85 #include "knowndrives.h"
89 // This is for solaris, where signal() resets the handler to SIG_DFL
90 // after the first signal is caught.
92 #define SIGNALFN sigset
94 #define SIGNALFN signal
98 #include "hostname_win32.h" // gethost/domainname()
99 #define HAVE_GETHOSTNAME 1
100 #define HAVE_GETDOMAINNAME 1
101 // fork()/signal()/initd simulation for native Windows
102 #include "daemon_win32.h" // daemon_main/detach/signal()
104 #define SIGNALFN daemon_signal
105 #define strsignal daemon_strsignal
106 #define sleep daemon_sleep
107 // SIGQUIT does not exist, CONTROL-Break signals SIGBREAK.
108 #define SIGQUIT SIGBREAK
109 #define SIGQUIT_KEYNAME "CONTROL-Break"
112 // 2x CONTROL-C simulates missing SIGQUIT via keyboard
113 #define SIGQUIT_KEYNAME "2x CONTROL-C"
115 #define SIGQUIT_KEYNAME "CONTROL-\\"
119 #if defined (__SVR4) && defined (__sun)
120 extern "C" int getdomainname(char *, int); // no declaration in header files!
123 #define ARGUSED(x) ((void)(x))
125 const char * smartd_cpp_cvsid
= "$Id: smartd.cpp 2984 2009-11-14 22:46:31Z chrfranke $"
126 CONFIG_H_CVSID EXTERN_H_CVSID
;
128 extern const char *reportbug
;
130 extern unsigned char debugmode
;
133 #define EXIT_BADCMD 1 // command line did not parse
134 #define EXIT_BADCONF 2 // syntax error in config file
135 #define EXIT_STARTUP 3 // problem forking daemon
136 #define EXIT_PID 4 // problem creating pid file
137 #define EXIT_NOCONF 5 // config file does not exist
138 #define EXIT_READCONF 6 // config file exists but cannot be read
140 #define EXIT_NOMEM 8 // out of memory
141 #define EXIT_BADCODE 10 // internal error - should NEVER happen
143 #define EXIT_BADDEV 16 // we can't monitor this device
144 #define EXIT_NODEV 17 // no devices to monitor
146 #define EXIT_SIGNAL 254 // abort on signal
148 // command-line: how long to sleep between checks
149 #define CHECKTIME 1800
150 static int checktime
=CHECKTIME
;
152 // command-line: name of PID file (empty for no pid file)
153 static std::string pid_file
;
155 // command-line: path prefix of persistent state file, empty if no persistence.
156 static std::string state_path_prefix
157 #ifdef SMARTMONTOOLS_SAVESTATES
158 = SMARTMONTOOLS_SAVESTATES
162 // command-line: path prefix of attribute log file, empty if no logs.
163 static std::string attrlog_path_prefix
164 #ifdef SMARTMONTOOLS_ATTRIBUTELOG
165 = SMARTMONTOOLS_ATTRIBUTELOG
169 // configuration file name
170 #define CONFIGFILENAME "smartd.conf"
173 static const char *configfile
= SMARTMONTOOLS_SYSCONFDIR
"/" CONFIGFILENAME
;
175 static const char *configfile
= "./" CONFIGFILENAME
;
177 // configuration file "name" if read from stdin
178 static const char * const configfile_stdin
= "<stdin>";
179 // path of alternate configuration file
180 static std::string configfile_alt
;
182 // command-line: when should we exit?
185 // command-line; this is the default syslog(3) log facility to use.
186 static int facility
=LOG_DAEMON
;
189 // command-line: fork into background?
190 static bool do_fork
=true;
193 // used for control of printing, passing arguments to atacmds.c
194 smartmonctrl
*con
=NULL
;
196 // set to one if we catch a USR1 (check devices now)
197 volatile int caughtsigUSR1
=0;
200 // set to one if we catch a USR2 (toggle debug mode)
201 volatile int caughtsigUSR2
=0;
204 // set to one if we catch a HUP (reload config file). In debug mode,
205 // set to two, if we catch INT (also reload config file).
206 volatile int caughtsigHUP
=0;
208 // set to signal value if we catch INT, QUIT, or TERM
209 volatile int caughtsigEXIT
=0;
211 // Attribute monitoring flags.
212 // See monitor_attr_flags below.
214 MONITOR_IGN_FAILUSE
= 0x01,
215 MONITOR_IGNORE
= 0x02,
216 MONITOR_RAW_PRINT
= 0x04,
218 MONITOR_AS_CRIT
= 0x10,
219 MONITOR_RAW_AS_CRIT
= 0x20,
222 // Array of flags for each attribute.
223 class attribute_flags
227 { memset(m_flags
, 0, sizeof(m_flags
)); }
229 bool is_set(int id
, unsigned char flag
) const
230 { return (0 < id
&& id
< (int)sizeof(m_flags
) && (m_flags
[id
] & flag
)); }
232 void set(int id
, unsigned char flags
)
234 if (0 < id
&& id
< (int)sizeof(m_flags
))
235 m_flags
[id
] |= flags
;
239 unsigned char m_flags
[256];
243 /// Configuration data for a device. Read from smartd.conf.
244 /// Supports copy & assignment and is compatible with STL containers.
247 int lineno
; // Line number of entry in file
248 std::string name
; // Device name
249 std::string dev_type
; // Device type argument from -d directive, empty if none
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 smartcheck
; // Check SMART status
253 bool usagefailed
; // Check for failed Usage Attributes
254 bool prefail
; // Track changes in Prefail Attributes
255 bool usage
; // Track changes in Usage Attributes
256 bool selftest
; // Monitor number of selftest errors
257 bool errorlog
; // Monitor number of ATA errors
258 bool permissive
; // Ignore failed SMART commands
259 char autosave
; // 1=disable, 2=enable Autosave Attributes
260 char autoofflinetest
; // 1=disable, 2=enable Auto Offline Test
261 unsigned char fix_firmwarebug
; // FIX_*, see atacmds.h
262 bool ignorepresets
; // Ignore database of -v options
263 bool showpresets
; // Show database entry for this device
264 bool removable
; // Device may disappear (not be present)
265 char powermode
; // skip check, if disk in idle or standby mode
266 bool powerquiet
; // skip powermode 'skipping checks' message
267 int powerskipmax
; // how many times can be check skipped
268 unsigned char tempdiff
; // Track Temperature changes >= this limit
269 unsigned char tempinfo
, tempcrit
; // Track Temperatures >= these limits as LOG_INFO, LOG_CRIT+mail
270 regular_expression test_regex
; // Regex for scheduled testing
272 // Configuration of email warning messages
273 std::string emailcmdline
; // script to execute, empty if no messages
274 std::string emailaddress
; // email address, or empty
275 unsigned char emailfreq
; // Emails once (1) daily (2) diminishing (3)
276 bool emailtest
; // Send test email?
279 unsigned char curr_pending_id
; // ID of current pending sector count, 0 if none
280 unsigned char offl_pending_id
; // ID of offline uncorrectable sector count, 0 if none
281 bool curr_pending_incr
, offl_pending_incr
; // True if current/offline pending values increase
282 bool curr_pending_set
, offl_pending_set
; // True if '-C', '-U' set in smartd.conf
284 attribute_flags monitor_attr_flags
; // MONITOR_* flags for each attribute
286 ata_vendor_attr_defs attribute_defs
; // -v options
291 dev_config::dev_config()
302 fix_firmwarebug(FIX_NOTSPECIFIED
),
303 ignorepresets(false),
310 tempinfo(0), tempcrit(0),
313 curr_pending_id(0), offl_pending_id(0),
314 curr_pending_incr(false), offl_pending_incr(false),
315 curr_pending_set(false), offl_pending_set(false)
320 // Number of allowed mail message types
321 const int SMARTD_NMAIL
= 13;
322 // Type for '-M test' mails (state not persistent)
323 const int MAILTYPE_TEST
= 0;
324 // TODO: Add const or enum for all mail types.
327 int logged
;// number of times an email has been sent
328 time_t firstsent
;// time first email was sent, as defined by time(2)
329 time_t lastsent
; // time last email was sent, as defined by time(2)
332 : logged(0), firstsent(0), lastsent(0) { }
335 /// Persistent state data for a device.
336 struct persistent_dev_state
338 unsigned char tempmin
, tempmax
; // Min/Max Temperatures
340 unsigned char selflogcount
; // total number of self-test errors
341 unsigned short selfloghour
; // lifetime hours of last self-test error
343 time_t scheduled_test_next_check
; // Time of next check for scheduled self-tests
345 mailinfo maillog
[SMARTD_NMAIL
]; // log info on when mail sent
348 int ataerrorcount
; // Total number of ATA errors
350 // Persistent part of ata_smart_values:
351 struct ata_attribute
{
354 unsigned char worst
; // Byte needed for 'raw64' attribute only.
357 ata_attribute() : id(0), val(0), worst(0), raw(0) { }
359 ata_attribute ata_attributes
[NUMBER_ATA_SMART_ATTRIBUTES
];
361 persistent_dev_state();
364 persistent_dev_state::persistent_dev_state()
365 : tempmin(0), tempmax(0),
368 scheduled_test_next_check(0),
373 /// Non-persistent state data for a device.
374 struct temp_dev_state
376 bool must_write
; // true if persistent part should be written
378 bool not_cap_offline
; // true == not capable of offline testing
379 bool not_cap_conveyance
;
382 bool not_cap_selective
;
384 unsigned char temperature
; // last recorded Temperature (in Celsius)
385 time_t tempmin_delay
; // time where Min Temperature tracking will start
387 bool powermodefail
; // true if power mode check failed
388 int powerskipcnt
; // Number of checks skipped due to idle or standby mode
391 unsigned char SmartPageSupported
; // has log sense IE page (0x2f)
392 unsigned char TempPageSupported
; // has log sense temperature page (0xd)
393 unsigned char SuppressReport
; // minimize nuisance reports
394 unsigned char modese_len
; // mode sense/select cmd len: 0 (don't
398 uint64_t num_sectors
; // Number of sectors (for selective self-test only)
399 ata_smart_values smartval
; // SMART data
400 ata_smart_thresholds_pvt smartthres
; // SMART thresholds
405 temp_dev_state::temp_dev_state()
407 not_cap_offline(false),
408 not_cap_conveyance(false),
409 not_cap_short(false),
411 not_cap_selective(false),
414 powermodefail(false),
416 SmartPageSupported(false),
417 TempPageSupported(false),
418 SuppressReport(false),
422 memset(&smartval
, 0, sizeof(smartval
));
423 memset(&smartthres
, 0, sizeof(smartthres
));
426 /// Runtime state data for a device.
428 : public persistent_dev_state
,
429 public temp_dev_state
431 void update_persistent_state();
432 void update_temp_state();
435 /// Container for configuration info for each device.
436 typedef std::vector
<dev_config
> dev_config_vector
;
438 /// Container for state info for each device.
439 typedef std::vector
<dev_state
> dev_state_vector
;
441 // Copy ATA attributes to persistent state.
442 void dev_state::update_persistent_state()
444 for (int i
= 0; i
< NUMBER_ATA_SMART_ATTRIBUTES
; i
++) {
445 const ata_smart_attribute
& ta
= smartval
.vendor_attributes
[i
];
446 ata_attribute
& pa
= ata_attributes
[i
];
449 pa
.val
= pa
.worst
= 0; pa
.raw
= 0;
457 | ((uint64_t)ta
.raw
[3] << 24)
458 | ((uint64_t)ta
.raw
[4] << 32)
459 | ((uint64_t)ta
.raw
[5] << 40);
463 // Copy ATA from persistent to temp state.
464 void dev_state::update_temp_state()
466 for (int i
= 0; i
< NUMBER_ATA_SMART_ATTRIBUTES
; i
++) {
467 const ata_attribute
& pa
= ata_attributes
[i
];
468 ata_smart_attribute
& ta
= smartval
.vendor_attributes
[i
];
471 ta
.current
= ta
.worst
= 0;
472 memset(ta
.raw
, 0, sizeof(ta
.raw
));
477 ta
.raw
[0] = (unsigned char) pa
.raw
;
478 ta
.raw
[1] = (unsigned char)(pa
.raw
>> 8);
479 ta
.raw
[2] = (unsigned char)(pa
.raw
>> 16);
480 ta
.raw
[3] = (unsigned char)(pa
.raw
>> 24);
481 ta
.raw
[4] = (unsigned char)(pa
.raw
>> 32);
482 ta
.raw
[5] = (unsigned char)(pa
.raw
>> 40);
486 // Parse a line from a state file.
487 static bool parse_dev_state_line(const char * line
, persistent_dev_state
& state
)
489 static regular_expression
regex(
491 "((temperature-min)" // (1 (2)
492 "|(temperature-max)" // (3)
493 "|(self-test-errors)" // (4)
494 "|(self-test-last-err-hour)" // (5)
495 "|(scheduled-test-next-check)" // (6)
496 "|(ata-error-count)" // (7)
497 "|(mail\\.([0-9]+)\\." // (8 (9)
498 "((count)" // (10 (11)
499 "|(first-sent-time)" // (12)
500 "|(last-sent-time)" // (13)
503 "|(ata-smart-attribute\\.([0-9]+)\\." // (14 (15)
511 " *= *([0-9]+)[ \n]*$", // (21)
515 throw std::logic_error("parse_dev_state_line: invalid regex");
517 const int nmatch
= 1+21;
518 regmatch_t match
[nmatch
];
519 if (!regex
.execute(line
, nmatch
, match
))
521 if (match
[nmatch
-1].rm_so
< 0)
524 uint64_t val
= strtoull(line
+ match
[nmatch
-1].rm_so
, (char **)0, 10);
527 if (match
[++m
].rm_so
>= 0)
528 state
.tempmin
= (unsigned char)val
;
529 else if (match
[++m
].rm_so
>= 0)
530 state
.tempmax
= (unsigned char)val
;
531 else if (match
[++m
].rm_so
>= 0)
532 state
.selflogcount
= (unsigned char)val
;
533 else if (match
[++m
].rm_so
>= 0)
534 state
.selfloghour
= (unsigned short)val
;
535 else if (match
[++m
].rm_so
>= 0)
536 state
.scheduled_test_next_check
= (time_t)val
;
537 else if (match
[++m
].rm_so
>= 0)
538 state
.ataerrorcount
= (int)val
;
539 else if (match
[m
+=2].rm_so
>= 0) {
540 int i
= atoi(line
+match
[m
].rm_so
);
541 if (!(0 <= i
&& i
< SMARTD_NMAIL
))
543 if (i
== MAILTYPE_TEST
) // Don't suppress test mails
545 if (match
[m
+=2].rm_so
>= 0)
546 state
.maillog
[i
].logged
= (int)val
;
547 else if (match
[++m
].rm_so
>= 0)
548 state
.maillog
[i
].firstsent
= (time_t)val
;
549 else if (match
[++m
].rm_so
>= 0)
550 state
.maillog
[i
].lastsent
= (time_t)val
;
554 else if (match
[m
+=5+1].rm_so
>= 0) {
555 int i
= atoi(line
+match
[m
].rm_so
);
556 if (!(0 <= i
&& i
< NUMBER_ATA_SMART_ATTRIBUTES
))
558 if (match
[m
+=2].rm_so
>= 0)
559 state
.ata_attributes
[i
].id
= (unsigned char)val
;
560 else if (match
[++m
].rm_so
>= 0)
561 state
.ata_attributes
[i
].val
= (unsigned char)val
;
562 else if (match
[++m
].rm_so
>= 0)
563 state
.ata_attributes
[i
].worst
= (unsigned char)val
;
564 else if (match
[++m
].rm_so
>= 0)
565 state
.ata_attributes
[i
].raw
= val
;
574 // Read a state file.
575 static bool read_dev_state(const char * path
, persistent_dev_state
& state
)
577 stdio_file
f(path
, "r");
580 pout("Cannot read state file \"%s\"\n", path
);
584 setmode(fileno(f
), O_TEXT
); // Allow files with \r\n
587 int good
= 0, bad
= 0;
589 while (fgets(line
, sizeof(line
), f
)) {
590 const char * s
= line
+ strspn(line
, " \t");
591 if (!*s
|| *s
== '#')
593 if (!parse_dev_state_line(line
, state
))
601 pout("%s: format error\n", path
);
604 pout("%s: %d invalid line(s) ignored\n", path
, bad
);
609 static void write_dev_state_line(FILE * f
, const char * name
, uint64_t val
)
612 fprintf(f
, "%s = %"PRIu64
"\n", name
, val
);
615 static void write_dev_state_line(FILE * f
, const char * name1
, int id
, const char * name2
, uint64_t val
)
618 fprintf(f
, "%s.%d.%s = %"PRIu64
"\n", name1
, id
, name2
, val
);
621 // Write a state file
622 static bool write_dev_state(const char * path
, const persistent_dev_state
& state
)
624 // Rename old "file" to "file~"
625 std::string pathbak
= path
; pathbak
+= '~';
626 unlink(pathbak
.c_str());
627 rename(path
, pathbak
.c_str());
629 stdio_file
f(path
, "w");
631 pout("Cannot create state file \"%s\"\n", path
);
635 fprintf(f
, "# smartd state file\n");
636 write_dev_state_line(f
, "temperature-min", state
.tempmin
);
637 write_dev_state_line(f
, "temperature-max", state
.tempmax
);
638 write_dev_state_line(f
, "self-test-errors", state
.selflogcount
);
639 write_dev_state_line(f
, "self-test-last-err-hour", state
.selfloghour
);
640 write_dev_state_line(f
, "scheduled-test-next-check", state
.scheduled_test_next_check
);
643 for (i
= 0; i
< SMARTD_NMAIL
; i
++) {
644 if (i
== MAILTYPE_TEST
) // Don't suppress test mails
646 const mailinfo
& mi
= state
.maillog
[i
];
649 write_dev_state_line(f
, "mail", i
, "count", mi
.logged
);
650 write_dev_state_line(f
, "mail", i
, "first-sent-time", mi
.firstsent
);
651 write_dev_state_line(f
, "mail", i
, "last-sent-time", mi
.lastsent
);
655 write_dev_state_line(f
, "ata-error-count", state
.ataerrorcount
);
657 for (i
= 0; i
< NUMBER_ATA_SMART_ATTRIBUTES
; i
++) {
658 const persistent_dev_state::ata_attribute
& pa
= state
.ata_attributes
[i
];
661 write_dev_state_line(f
, "ata-smart-attribute", i
, "id", pa
.id
);
662 write_dev_state_line(f
, "ata-smart-attribute", i
, "val", pa
.val
);
663 write_dev_state_line(f
, "ata-smart-attribute", i
, "worst", pa
.worst
);
664 write_dev_state_line(f
, "ata-smart-attribute", i
, "raw", pa
.raw
);
670 // Write to the attrlog file
671 static bool write_dev_attrlog(const char * path
, const persistent_dev_state
& state
)
673 stdio_file
f(path
, "a");
675 pout("Cannot create attribute log file \"%s\"\n", path
);
680 time_t now
= time(0);
681 struct tm
* tms
= gmtime(&now
);
682 fprintf(f
, "%d-%02d-%02d %02d:%02d:%02d;",
683 1900+tms
->tm_year
, 1+tms
->tm_mon
, tms
->tm_mday
,
684 tms
->tm_hour
, tms
->tm_min
, tms
->tm_sec
);
685 for (int i
= 0; i
< NUMBER_ATA_SMART_ATTRIBUTES
; i
++) {
686 const persistent_dev_state::ata_attribute
& pa
= state
.ata_attributes
[i
];
689 fprintf(f
, "\t%d;%d;%"PRIu64
";", pa
.id
, pa
.val
, pa
.raw
);
696 // Write all state files. If write_always is false, don't write
697 // unless must_write is set.
698 static void write_all_dev_states(const dev_config_vector
& configs
,
699 dev_state_vector
& states
,
700 bool write_always
= true)
702 for (unsigned i
= 0; i
< states
.size(); i
++) {
703 const dev_config
& cfg
= configs
.at(i
);
704 if (cfg
.state_file
.empty())
706 dev_state
& state
= states
[i
];
707 if (!write_always
&& !state
.must_write
)
709 if (!write_dev_state(cfg
.state_file
.c_str(), state
))
711 state
.must_write
= false;
712 if (write_always
|| debugmode
)
713 PrintOut(LOG_INFO
, "Device: %s, state written to %s\n",
714 cfg
.name
.c_str(), cfg
.state_file
.c_str());
718 // Write to all attrlog files
719 static void write_all_dev_attrlogs(const dev_config_vector
& configs
,
720 dev_state_vector
& states
)
722 for (unsigned i
= 0; i
< states
.size(); i
++) {
723 const dev_config
& cfg
= configs
.at(i
);
724 if (cfg
.attrlog_file
.empty())
726 dev_state
& state
= states
[i
];
727 write_dev_attrlog(cfg
.attrlog_file
.c_str(), state
);
731 // remove the PID file
732 void RemovePidFile(){
733 if (!pid_file
.empty()) {
734 if (unlink(pid_file
.c_str()))
735 PrintOut(LOG_CRIT
,"Can't unlink PID file %s (%s).\n",
736 pid_file
.c_str(), strerror(errno
));
742 extern "C" { // signal handlers require C-linkage
744 // Note if we catch a SIGUSR1
745 void USR1handler(int sig
){
752 // Note if we catch a SIGUSR2
753 void USR2handler(int sig
){
760 // Note if we catch a HUP (or INT in debug mode)
761 void HUPhandler(int sig
){
769 // signal handler for TERM, QUIT, and INT (if not in debug mode)
770 void sighandler(int sig
){
778 // Cleanup, print Goodbye message and remove pidfile
779 static int Goodbye(int status
)
781 // delete PID file, if one was created
784 // if we are exiting because of a code bug, tell user
785 if (status
==EXIT_BADCODE
)
786 PrintOut(LOG_CRIT
, "Please inform " PACKAGE_BUGREPORT
", including output of smartd -V.\n");
788 // and this should be the final output from smartd before it exits
789 PrintOut(status
?LOG_CRIT
:LOG_INFO
, "smartd is exiting (exit status %d)\n", status
);
794 #define ENVLENGTH 1024
796 // a replacement for setenv() which is not available on all platforms.
797 // Note that the string passed to putenv must not be freed or made
798 // invalid, since a pointer to it is kept by putenv(). This means that
799 // it must either be a static buffer or allocated off the heap. The
800 // string can be freed if the environment variable is redefined or
801 // deleted via another call to putenv(). So we keep these on the stack
802 // as long as the popen() call is underway.
803 int exportenv(char* stackspace
, const char *name
, const char *value
){
804 snprintf(stackspace
,ENVLENGTH
, "%s=%s", name
, value
);
805 return putenv(stackspace
);
808 char* dnsdomain(const char* hostname
) {
810 #ifdef HAVE_GETADDRINFO
811 static char canon_name
[NI_MAXHOST
];
812 struct addrinfo
*info
= NULL
;
813 struct addrinfo hints
;
816 memset(&hints
, 0, sizeof(hints
));
817 hints
.ai_flags
= AI_CANONNAME
;
818 if ((err
= getaddrinfo(hostname
, NULL
, &hints
, &info
)) || (!info
)) {
819 PrintOut(LOG_CRIT
, "Error retrieving getaddrinfo(%s): %s\n", hostname
, gai_strerror(err
));
822 if (info
->ai_canonname
) {
823 strncpy(canon_name
, info
->ai_canonname
, sizeof(canon_name
));
824 canon_name
[NI_MAXHOST
- 1] = '\0';
826 if ((p
= strchr(canon_name
, '.')))
830 #elif HAVE_GETHOSTBYNAME
832 if ((hp
= gethostbyname(hostname
))) {
833 // Does this work if gethostbyname() returns an IPv6 name in
834 // colon/dot notation? [BA]
835 if ((p
= strchr(hp
->h_name
, '.')))
846 static void MailWarning(const dev_config
& cfg
, dev_state
& state
, int which
, const char *fmt
, ...)
847 __attribute__ ((format (printf
, 4, 5)));
849 // If either address or executable path is non-null then send and log
850 // a warning email, or execute executable
851 static void MailWarning(const dev_config
& cfg
, dev_state
& state
, int which
, const char *fmt
, ...){
852 char command
[2048], message
[256], hostname
[256], domainname
[256], additional
[256],fullmessage
[1024];
853 char original
[256], further
[256], nisdomain
[256], subject
[256],dates
[DATEANDEPOCHLEN
];
854 char environ_strings
[11][ENVLENGTH
];
857 const int day
=24*3600;
859 const char * const whichfail
[]={
865 "FailedHealthCheck", // 5
866 "FailedReadSmartData", // 6
867 "FailedReadSmartErrorLog", // 7
868 "FailedReadSmartSelfTestLog", // 8
869 "FailedOpenDevice", // 9
870 "CurrentPendingSector", // 10
871 "OfflineUncorrectableSector", // 11
875 const char *unknown
="[Unknown]";
877 // See if user wants us to send mail
878 if (cfg
.emailaddress
.empty() && cfg
.emailcmdline
.empty())
881 std::string address
= cfg
.emailaddress
;
882 const char * executable
= cfg
.emailcmdline
.c_str();
884 // which type of mail are we sending?
885 mailinfo
* mail
=(state
.maillog
)+which
;
888 if (cfg
.emailfreq
<1 || cfg
.emailfreq
>3) {
889 PrintOut(LOG_CRIT
,"internal error in MailWarning(): cfg.mailwarn->emailfreq=%d\n",cfg
.emailfreq
);
892 if (which
<0 || which
>=SMARTD_NMAIL
|| sizeof(whichfail
)!=SMARTD_NMAIL
*sizeof(char *)) {
893 PrintOut(LOG_CRIT
,"Contact " PACKAGE_BUGREPORT
"; internal error in MailWarning(): which=%d, size=%d\n",
894 which
, (int)sizeof(whichfail
));
898 // Return if a single warning mail has been sent.
899 if ((cfg
.emailfreq
==1) && mail
->logged
)
902 // Return if this is an email test and one has already been sent.
903 if (which
== 0 && mail
->logged
)
906 // To decide if to send mail, we need to know what time it is.
909 // Return if less than one day has gone by
910 if (cfg
.emailfreq
==2 && mail
->logged
&& epoch
<(mail
->lastsent
+day
))
913 // Return if less than 2^(logged-1) days have gone by
914 if (cfg
.emailfreq
==3 && mail
->logged
) {
915 days
=0x01<<(mail
->logged
-1);
917 if (epoch
<(mail
->lastsent
+days
))
921 // record the time of this mail message, and the first mail message
923 mail
->firstsent
=epoch
;
924 mail
->lastsent
=epoch
;
926 // get system host & domain names (not null terminated if length=MAX)
927 #ifdef HAVE_GETHOSTNAME
928 if (gethostname(hostname
, 256))
929 strcpy(hostname
, unknown
);
933 p
= dnsdomain(hostname
);
935 strncpy(domainname
, p
, 255);
936 domainname
[255]='\0';
938 strcpy(domainname
, unknown
);
941 strcpy(hostname
, unknown
);
942 strcpy(domainname
, unknown
);
945 #ifdef HAVE_GETDOMAINNAME
946 if (getdomainname(nisdomain
, 256))
947 strcpy(nisdomain
, unknown
);
951 strcpy(nisdomain
, unknown
);
954 // print warning string into message
956 vsnprintf(message
, 256, fmt
, ap
);
959 // appropriate message about further information
960 additional
[0]=original
[0]=further
[0]='\0';
962 sprintf(further
,"You can also use the smartctl utility for further investigation.\n");
964 switch (cfg
.emailfreq
) {
966 sprintf(additional
,"No additional email messages about this problem will be sent.\n");
969 sprintf(additional
,"Another email message will be sent in 24 hours if the problem persists.\n");
972 sprintf(additional
,"Another email message will be sent in %d days if the problem persists\n",
973 (0x01)<<mail
->logged
);
976 if (cfg
.emailfreq
>1 && mail
->logged
) {
977 dateandtimezoneepoch(dates
, mail
->firstsent
);
978 sprintf(original
,"The original email about this issue was sent at %s\n", dates
);
982 snprintf(subject
, 256,"SMART error (%s) detected on host: %s", whichfail
[which
], hostname
);
984 // If the user has set cfg.emailcmdline, use that as mailer, else "mail" or "mailx".
986 #ifdef DEFAULT_MAILER
987 executable
= DEFAULT_MAILER
;
992 executable
= "blat"; // http://blat.sourceforge.net/
996 #ifndef _WIN32 // blat mailer needs comma
997 // replace commas by spaces to separate recipients
998 std::replace(address
.begin(), address
.end(), ',', ' ');
1000 // Export information in environment variables that will be useful
1002 exportenv(environ_strings
[0], "SMARTD_MAILER", executable
);
1003 exportenv(environ_strings
[1], "SMARTD_MESSAGE", message
);
1004 exportenv(environ_strings
[2], "SMARTD_SUBJECT", subject
);
1005 dateandtimezoneepoch(dates
, mail
->firstsent
);
1006 exportenv(environ_strings
[3], "SMARTD_TFIRST", dates
);
1007 snprintf(dates
, DATEANDEPOCHLEN
,"%d", (int)mail
->firstsent
);
1008 exportenv(environ_strings
[4], "SMARTD_TFIRSTEPOCH", dates
);
1009 exportenv(environ_strings
[5], "SMARTD_FAILTYPE", whichfail
[which
]);
1010 if (!address
.empty())
1011 exportenv(environ_strings
[6], "SMARTD_ADDRESS", address
.c_str());
1012 exportenv(environ_strings
[7], "SMARTD_DEVICESTRING", cfg
.name
.c_str());
1014 exportenv(environ_strings
[8], "SMARTD_DEVICETYPE", cfg
.dev_type
.c_str());
1015 exportenv(environ_strings
[9], "SMARTD_DEVICE", cfg
.name
.c_str());
1017 snprintf(fullmessage
, 1024,
1018 "This email was generated by the smartd daemon running on:\n\n"
1021 " NIS domain: %s\n\n"
1022 "The following warning/error was logged by the smartd daemon:\n\n"
1024 "For details see host's SYSLOG (default: /var/log/messages).\n\n"
1026 hostname
, domainname
, nisdomain
, message
, further
, original
, additional
);
1027 exportenv(environ_strings
[10], "SMARTD_FULLMESSAGE", fullmessage
);
1029 // now construct a command to send this as EMAIL
1031 if (!address
.empty())
1032 snprintf(command
, 2048,
1033 "$SMARTD_MAILER -s '%s' %s 2>&1 << \"ENDMAIL\"\n"
1034 "%sENDMAIL\n", subject
, address
.c_str(), fullmessage
);
1036 snprintf(command
, 2048, "%s 2>&1", executable
);
1038 // tell SYSLOG what we are about to do...
1039 const char * newadd
= (!address
.empty()? address
.c_str() : "<nomailer>");
1040 const char * newwarn
= (which
? "Warning via" : "Test of");
1042 PrintOut(LOG_INFO
,"%s %s to %s ...\n",
1043 which
?"Sending warning via":"Executing test of", executable
, newadd
);
1045 // issue the command to send mail or to run the user's executable
1048 if (!(pfp
=popen(command
, "r")))
1049 // failed to popen() mail process
1050 PrintOut(LOG_CRIT
,"%s %s to %s: failed (fork or pipe failed, or no memory) %s\n",
1051 newwarn
, executable
, newadd
, errno
?strerror(errno
):"");
1055 char buffer
[EBUFLEN
];
1057 // if unexpected output on stdout/stderr, null terminate, print, and flush
1058 if ((len
=fread(buffer
, 1, EBUFLEN
, pfp
))) {
1060 int newlen
= len
<EBUFLEN
? len
: EBUFLEN
-1;
1061 buffer
[newlen
]='\0';
1062 PrintOut(LOG_CRIT
,"%s %s to %s produced unexpected output (%s%d bytes) to STDOUT/STDERR: \n%s\n",
1063 newwarn
, executable
, newadd
, len
!=newlen
?"here truncated to ":"", newlen
, buffer
);
1065 // flush pipe if needed
1066 while (fread(buffer
, 1, EBUFLEN
, pfp
) && count
<EBUFLEN
)
1069 // tell user that pipe was flushed, or that something is really wrong
1070 if (count
&& count
<EBUFLEN
)
1071 PrintOut(LOG_CRIT
,"%s %s to %s: flushed remaining STDOUT/STDERR\n",
1072 newwarn
, executable
, newadd
);
1074 PrintOut(LOG_CRIT
,"%s %s to %s: more than 1 MB STDOUT/STDERR flushed, breaking pipe\n",
1075 newwarn
, executable
, newadd
);
1078 // if something went wrong with mail process, print warning
1080 if (-1==(status
=pclose(pfp
)))
1081 PrintOut(LOG_CRIT
,"%s %s to %s: pclose(3) failed %s\n", newwarn
, executable
, newadd
,
1082 errno
?strerror(errno
):"");
1084 // mail process apparently succeeded. Check and report exit status
1087 if (WIFEXITED(status
)) {
1088 // exited 'normally' (but perhaps with nonzero status)
1089 status8
=WEXITSTATUS(status
);
1092 PrintOut(LOG_CRIT
,"%s %s to %s: failed (32-bit/8-bit exit status: %d/%d) perhaps caught signal %d [%s]\n",
1093 newwarn
, executable
, newadd
, status
, status8
, status8
-128, strsignal(status8
-128));
1095 PrintOut(LOG_CRIT
,"%s %s to %s: failed (32-bit/8-bit exit status: %d/%d)\n",
1096 newwarn
, executable
, newadd
, status
, status8
);
1098 PrintOut(LOG_INFO
,"%s %s to %s: successful\n", newwarn
, executable
, newadd
);
1101 if (WIFSIGNALED(status
))
1102 PrintOut(LOG_INFO
,"%s %s to %s: exited because of uncaught signal %d [%s]\n",
1103 newwarn
, executable
, newadd
, WTERMSIG(status
), strsignal(WTERMSIG(status
)));
1105 // this branch is probably not possible. If subprocess is
1106 // stopped then pclose() should not return.
1107 if (WIFSTOPPED(status
))
1108 PrintOut(LOG_CRIT
,"%s %s to %s: process STOPPED because it caught signal %d [%s]\n",
1109 newwarn
, executable
, newadd
, WSTOPSIG(status
), strsignal(WSTOPSIG(status
)));
1116 // No "here-documents" on Windows, so must use separate commandline and stdin
1117 char stdinbuf
[1024];
1118 command
[0] = stdinbuf
[0] = 0;
1119 int boxtype
= -1, boxmsgoffs
= 0;
1120 const char * newadd
= "<nomailer>";
1121 if (!address
.empty()) {
1122 // address "[sys]msgbox ..." => show warning (also) as [system modal ]messagebox
1123 char addr1
[9+1+13] = ""; int n1
= -1, n2
= -1;
1124 if (sscanf(address
.c_str(), "%9[a-z]%n,%n", addr1
, &n1
, &n2
) == 1 && (n1
== (int)address
.size() || n2
> 0)) {
1125 if (!strcmp(addr1
, "msgbox"))
1127 else if (!strcmp(addr1
, "sysmsgbox"))
1130 address
.erase(0, (n2
> n1
? n2
: n1
));
1133 if (!address
.empty()) {
1134 // Use "blat" parameter syntax (TODO: configure via -M for other mailers)
1135 snprintf(command
, sizeof(command
),
1136 "%s - -q -subject \"%s\" -to \"%s\"",
1137 executable
, subject
, address
.c_str());
1138 newadd
= address
.c_str();
1142 _set_printf_count_output(1); // "%n" disabled by default
1144 // Message for mail [0...] and messagebox [boxmsgoffs...]
1145 snprintf(stdinbuf
, sizeof(stdinbuf
),
1146 "This email was generated by the smartd daemon running on:\n\n"
1149 // " NIS domain: %s\n"
1151 "The following warning/error was logged by the smartd daemon:\n\n"
1153 "For details see the event log or log file of smartd.\n\n"
1156 hostname
, /*domainname, */ nisdomain
, &boxmsgoffs
, message
, further
, original
, additional
);
1159 snprintf(command
, sizeof(command
), "%s", executable
);
1161 const char * newwarn
= (which
? "Warning via" : "Test of");
1164 daemon_messagebox(boxtype
, subject
, stdinbuf
+boxmsgoffs
);
1165 PrintOut(LOG_INFO
,"%s message box\n", newwarn
);
1168 char stdoutbuf
[800]; // < buffer in syslog_win32::vsyslog()
1171 PrintOut(LOG_INFO
,"%s %s to %s ...\n",
1172 (which
?"Sending warning via":"Executing test of"), executable
, newadd
);
1173 rc
= daemon_spawn(command
, stdinbuf
, strlen(stdinbuf
), stdoutbuf
, sizeof(stdoutbuf
));
1174 if (rc
>= 0 && stdoutbuf
[0])
1175 PrintOut(LOG_CRIT
,"%s %s to %s produced unexpected output (%d bytes) to STDOUT/STDERR:\n%s\n",
1176 newwarn
, executable
, newadd
, strlen(stdoutbuf
), stdoutbuf
);
1178 PrintOut(LOG_CRIT
,"%s %s to %s: failed, exit status %d\n",
1179 newwarn
, executable
, newadd
, rc
);
1181 PrintOut(LOG_INFO
,"%s %s to %s: successful\n", newwarn
, executable
, newadd
);
1186 // increment mail sent counter
1190 // Printing function for watching ataprint commands, or losing them
1191 // [From GLIBC Manual: Since the prototype doesn't specify types for
1192 // optional arguments, in a call to a variadic function the default
1193 // argument promotions are performed on the optional argument
1194 // values. This means the objects of type char or short int (whether
1195 // signed or not) are promoted to either int or unsigned int, as
1197 void pout(const char *fmt
, ...){
1200 // get the correct time in syslog()
1201 FixGlibcTimeZoneBug();
1202 // initialize variable argument list
1204 // in debug==1 mode we will print the output from the ataprint.o functions!
1205 if (debugmode
&& debugmode
!=2)
1207 if (facility
== LOG_LOCAL1
) // logging to stdout
1208 vfprintf(stderr
,fmt
,ap
);
1212 // in debug==2 mode we print output from knowndrives.o functions
1213 else if (debugmode
==2 || con
->reportataioctl
|| con
->reportscsiioctl
/*|| con->controller_port???*/) {
1214 openlog("smartd", LOG_PID
, facility
);
1215 vsyslog(LOG_INFO
, fmt
, ap
);
1223 // This function prints either to stdout or to the syslog as needed.
1224 // This function is also used by utility.cpp to report LOG_CRIT errors.
1225 void PrintOut(int priority
, const char *fmt
, ...){
1228 // get the correct time in syslog()
1229 FixGlibcTimeZoneBug();
1230 // initialize variable argument list
1234 if (facility
== LOG_LOCAL1
) // logging to stdout
1235 vfprintf(stderr
,fmt
,ap
);
1240 openlog("smartd", LOG_PID
, facility
);
1241 vsyslog(priority
,fmt
,ap
);
1248 // Used to warn users about invalid checksums. Called from atacmds.cpp.
1249 void checksumwarning(const char * string
)
1251 pout("Warning! %s error: invalid SMART checksum.\n", string
);
1254 // Wait for the pid file to show up, this makes sure a calling program knows
1255 // that the daemon is really up and running and has a pid to kill it
1256 bool WaitForPidFile()
1258 int waited
, max_wait
= 10;
1259 struct stat stat_buf
;
1261 if (pid_file
.empty() || debugmode
)
1264 for(waited
= 0; waited
< max_wait
; ++waited
) {
1265 if (!stat(pid_file
.c_str(), &stat_buf
)) {
1274 // Forks new process, closes ALL file descriptors, redirects stdin,
1275 // stdout, and stderr. Not quite daemon(). See
1276 // http://www.linuxjournal.com/article/2335
1277 // for a good description of why we do things this way.
1283 // flush all buffered streams. Else we might get two copies of open
1284 // streams since both parent and child get copies of the buffers.
1288 if ((pid
=fork()) < 0) {
1290 PrintOut(LOG_CRIT
,"smartd unable to fork daemon process!\n");
1294 // we are the parent process, wait for pid file, then exit cleanly
1295 if(!WaitForPidFile()) {
1296 PrintOut(LOG_CRIT
,"PID file %s didn't show up!\n", pid_file
.c_str());
1302 // from here on, we are the child process.
1305 // Fork one more time to avoid any possibility of having terminals
1306 if ((pid
=fork()) < 0) {
1308 PrintOut(LOG_CRIT
,"smartd unable to fork daemon process!\n");
1312 // we are the parent process -- exit cleanly
1315 // Now we are the child's child...
1318 // close any open file descriptors
1319 for (i
=getdtablesize();i
>=0;--i
)
1323 // Cygwin's setsid() does not detach the process from Windows console
1325 #endif // __CYGWIN__
1327 #define NO_warn_unused_result(cmd) { if (cmd) {} ; }
1329 // redirect any IO attempts to /dev/null for stdin
1330 i
=open("/dev/null",O_RDWR
);
1333 NO_warn_unused_result(dup(i
));
1335 NO_warn_unused_result(dup(i
));
1338 NO_warn_unused_result(chdir("/"));
1341 PrintOut(LOG_INFO
, "smartd has fork()ed into background mode. New PID=%d.\n", (int)getpid());
1345 // No fork() on native Win32
1346 // Detach this process from console
1348 if (daemon_detach("smartd")) {
1349 PrintOut(LOG_CRIT
,"smartd unable to detach from console!\n");
1352 // stdin/out/err now closed if not redirected
1358 // create a PID file containing the current process id
1359 static void WritePidFile()
1361 if (!pid_file
.empty()) {
1362 pid_t pid
= getpid();
1365 old_umask
= umask(0077); // rwx------
1367 // Cygwin: smartd service runs on system account, ensure PID file can be read by admins
1368 old_umask
= umask(0033); // rwxr--r--
1371 stdio_file
f(pid_file
.c_str(), "w");
1373 if (!(f
&& fprintf(f
, "%d\n", (int)pid
) > 0 && f
.close())) {
1374 PrintOut(LOG_CRIT
, "unable to write PID file %s - exiting.\n", pid_file
.c_str());
1377 PrintOut(LOG_INFO
, "file %s written containing PID %d\n", pid_file
.c_str(), (int)pid
);
1381 // Prints header identifying version of code and home
1382 static void PrintHead()
1384 PrintOut(LOG_INFO
, "%s\n", format_version_info("smartd").c_str());
1387 // prints help info for configuration file Directives
1390 "Configuration file (%s) Directives (after device name):\n"
1391 " -d TYPE Set the device type: %s\n"
1392 " -T TYPE Set the tolerance to one of: normal, permissive\n"
1393 " -o VAL Enable/disable automatic offline tests (on/off)\n"
1394 " -S VAL Enable/disable attribute autosave (on/off)\n"
1395 " -n MODE No check if: never, sleep[,N][,q], standby[,N][,q], idle[,N][,q]\n"
1396 " -H Monitor SMART Health Status, report if failed\n"
1397 " -s REG Do Self-Test at time(s) given by regular expression REG\n"
1398 " -l TYPE Monitor SMART log. Type is one of: error, selftest\n"
1399 " -f Monitor 'Usage' Attributes, report failures\n"
1400 " -m ADD Send email warning to address ADD\n"
1401 " -M TYPE Modify email warning behavior (see man page)\n"
1402 " -p Report changes in 'Prefailure' Attributes\n"
1403 " -u Report changes in 'Usage' Attributes\n"
1404 " -t Equivalent to -p and -u Directives\n"
1405 " -r ID Also report Raw values of Attribute ID with -p, -u or -t\n"
1406 " -R ID Track changes in Attribute ID Raw value with -p, -u or -t\n"
1407 " -i ID Ignore Attribute ID for -f Directive\n"
1408 " -I ID Ignore Attribute ID for -p, -u or -t Directive\n"
1409 " -C ID[+] Monitor [increases of] Current Pending Sectors in Attribute ID\n"
1410 " -U ID[+] Monitor [increases of] Offline Uncorrectable Sectors in Attribute ID\n"
1411 " -W D,I,C Monitor Temperature D)ifference, I)nformal limit, C)ritical limit\n"
1412 " -v N,ST Modifies labeling of Attribute N (see man page) \n"
1413 " -P TYPE Drive-specific presets: use, ignore, show, showall\n"
1414 " -a Default: -H -f -t -l error -l selftest -C 197 -U 198\n"
1415 " -F TYPE Firmware bug workaround: none, samsung, samsung2, samsung3\n"
1416 " # Comment: text after a hash sign is ignored\n"
1417 " \\ Line continuation character\n"
1418 "Attribute ID is a decimal integer 1 <= ID <= 255\n"
1419 "Use ID = 0 to turn off -C and/or -U Directives\n"
1420 "Example: /dev/hda -a\n",
1421 configfile
, smi()->get_valid_dev_types_str().c_str());
1425 /* Returns a pointer to a static string containing a formatted list of the valid
1426 arguments to the option opt or NULL on failure. */
1427 const char *GetValidArgList(char opt
) {
1431 return "<PATH_PREFIX>";
1433 return "<FILE_NAME>, -";
1435 return "daemon, local0, local1, local2, local3, local4, local5, local6, local7";
1437 return "nodev, errors, nodevstartup, never, onecheck, showtests";
1439 return "ioctl[,N], ataioctl[,N], scsiioctl[,N]";
1442 return "<FILE_NAME>";
1444 return "<INTEGER_SECONDS>";
1450 /* prints help information for command syntax */
1452 PrintOut(LOG_INFO
,"Usage: smartd [options]\n\n");
1453 PrintOut(LOG_INFO
," -A PREFIX, --attributelog=PREFIX\n");
1454 PrintOut(LOG_INFO
," Log ATA attribute information to {PREFIX}MODEL-SERIAL.ata.csv\n");
1455 #ifdef SMARTMONTOOLS_ATTRIBUTELOG
1456 PrintOut(LOG_INFO
," [default is "SMARTMONTOOLS_ATTRIBUTELOG
"MODEL-SERIAL.ata.csv]\n");
1458 PrintOut(LOG_INFO
,"\n");
1459 PrintOut(LOG_INFO
," -B [+]FILE, --drivedb=[+]FILE\n");
1460 PrintOut(LOG_INFO
," Read and replace [add] drive database from FILE\n");
1461 #ifdef SMARTMONTOOLS_DRIVEDBDIR
1462 PrintOut(LOG_INFO
," [default is "SMARTMONTOOLS_DRIVEDBDIR
"/drivedb.h]\n");
1464 PrintOut(LOG_INFO
,"\n");
1465 PrintOut(LOG_INFO
," -c NAME|-, --configfile=NAME|-\n");
1466 PrintOut(LOG_INFO
," Read configuration file NAME or stdin [default is %s]\n\n", configfile
);
1467 PrintOut(LOG_INFO
," -d, --debug\n");
1468 PrintOut(LOG_INFO
," Start smartd in debug mode\n\n");
1469 PrintOut(LOG_INFO
," -D, --showdirectives\n");
1470 PrintOut(LOG_INFO
," Print the configuration file Directives and exit\n\n");
1471 PrintOut(LOG_INFO
," -h, --help, --usage\n");
1472 PrintOut(LOG_INFO
," Display this help and exit\n\n");
1473 PrintOut(LOG_INFO
," -i N, --interval=N\n");
1474 PrintOut(LOG_INFO
," Set interval between disk checks to N seconds, where N >= 10\n\n");
1475 PrintOut(LOG_INFO
," -l local[0-7], --logfacility=local[0-7]\n");
1477 PrintOut(LOG_INFO
," Use syslog facility local0 - local7 or daemon [default]\n\n");
1479 PrintOut(LOG_INFO
," Log to \"./smartd.log\", stdout, stderr [default is event log]\n\n");
1482 PrintOut(LOG_INFO
," -n, --no-fork\n");
1483 PrintOut(LOG_INFO
," Do not fork into background\n\n");
1485 PrintOut(LOG_INFO
," -p NAME, --pidfile=NAME\n");
1486 PrintOut(LOG_INFO
," Write PID file NAME\n\n");
1487 PrintOut(LOG_INFO
," -q WHEN, --quit=WHEN\n");
1488 PrintOut(LOG_INFO
," Quit on one of: %s\n\n", GetValidArgList('q'));
1489 PrintOut(LOG_INFO
," -r, --report=TYPE\n");
1490 PrintOut(LOG_INFO
," Report transactions for one of: %s\n\n", GetValidArgList('r'));
1491 PrintOut(LOG_INFO
," -s PREFIX, --savestates=PREFIX\n");
1492 PrintOut(LOG_INFO
," Save disk states to {PREFIX}MODEL-SERIAL.TYPE.state\n");
1493 #ifdef SMARTMONTOOLS_SAVESTATES
1494 PrintOut(LOG_INFO
," [default is "SMARTMONTOOLS_SAVESTATES
"MODEL-SERIAL.TYPE.state]\n");
1496 PrintOut(LOG_INFO
,"\n");
1498 PrintOut(LOG_INFO
," --service\n");
1499 PrintOut(LOG_INFO
," Running as windows service (see man page), install with:\n");
1500 PrintOut(LOG_INFO
," smartd install [options]\n");
1501 PrintOut(LOG_INFO
," Remove service with:\n");
1502 PrintOut(LOG_INFO
," smartd remove\n\n");
1504 PrintOut(LOG_INFO
," -V, --version, --license, --copyright\n");
1505 PrintOut(LOG_INFO
," Print License, Copyright, and version information\n");
1508 static int CloseDevice(smart_device
* device
, const char * name
)
1510 if (!device
->close()){
1511 PrintOut(LOG_INFO
,"Device: %s, %s, close() failed\n", name
, device
->get_errmsg());
1514 // device sucessfully closed
1518 // return true if a char is not allowed in a state file name
1519 static bool not_allowed_in_filename(char c
)
1521 return !( ('0' <= c
&& c
<= '9')
1522 || ('A' <= c
&& c
<= 'Z')
1523 || ('a' <= c
&& c
<= 'z'));
1526 // returns <0 on failure
1527 static int ATAErrorCount(ata_device
* device
, const char * name
,
1528 unsigned char fix_firmwarebug
)
1530 struct ata_smart_errorlog log
;
1532 if (ataReadErrorLog(device
, &log
, fix_firmwarebug
)){
1533 PrintOut(LOG_INFO
,"Device: %s, Read SMART Error Log Failed\n",name
);
1537 // return current number of ATA errors
1538 return log
.error_log_pointer
?log
.ata_error_count
:0;
1541 // returns <0 if problem. Otherwise, bottom 8 bits are the self test
1542 // error count, and top bits are the power-on hours of the last error.
1543 static int SelfTestErrorCount(ata_device
* device
, const char * name
,
1544 unsigned char fix_firmwarebug
)
1546 struct ata_smart_selftestlog log
;
1548 if (ataReadSelfTestLog(device
, &log
, fix_firmwarebug
)){
1549 PrintOut(LOG_INFO
,"Device: %s, Read SMART Self Test Log Failed\n",name
);
1553 // return current number of self-test errors
1554 return ataPrintSmartSelfTestlog(&log
, false, fix_firmwarebug
);
1557 #define SELFTEST_ERRORCOUNT(x) (x & 0xff)
1558 #define SELFTEST_ERRORHOURS(x) ((x >> 8) & 0xffff)
1560 // Log self-test execution status
1561 static void log_self_test_exec_status(const char * name
, unsigned char status
)
1564 switch (status
>> 4) {
1565 case 0x0: msg
= "completed without error"; break;
1566 case 0x1: msg
= "was aborted by the host"; break;
1567 case 0x2: msg
= "was interrupted by the host with a reset"; break;
1568 case 0x3: msg
= "could not complete due to a fatal or unknown error"; break;
1569 case 0x4: msg
= "completed with error (unknown test element)"; break;
1570 case 0x5: msg
= "completed with error (electrical test element)"; break;
1571 case 0x6: msg
= "completed with error (servo/seek test element)"; break;
1572 case 0x7: msg
= "completed with error (read test element)"; break;
1573 case 0x8: msg
= "completed with error (handling damage?)"; break;
1578 PrintOut(((status
>> 4) >= 0x4 ? LOG_CRIT
: LOG_INFO
),
1579 "Device: %s, previous self-test %s\n", name
, msg
);
1580 else if ((status
>> 4) == 0xf)
1581 PrintOut(LOG_INFO
, "Device: %s, self-test in progress, %u0%% remaining\n",
1582 name
, status
& 0x0f);
1584 PrintOut(LOG_INFO
, "Device: %s, unknown self-test status 0x%02x\n",
1589 // TODO: Add '-F swapid' directive
1590 const bool fix_swapped_id
= false;
1592 // scan to see what ata devices there are, and if they support SMART
1593 static int ATADeviceScan(dev_config
& cfg
, dev_state
& state
, ata_device
* atadev
)
1596 struct ata_identify_device drive
;
1597 const char *name
= cfg
.name
.c_str();
1600 // Device must be open
1602 // Get drive identity structure
1603 if ((retid
=ataReadHDIdentity (atadev
, &drive
))){
1605 // Unable to read Identity structure
1606 PrintOut(LOG_INFO
,"Device: %s, not ATA, no IDENTIFY DEVICE Structure\n",name
);
1608 PrintOut(LOG_INFO
,"Device: %s, packet devices [this device %s] not SMART capable\n",
1609 name
, packetdevicetype(retid
-1));
1610 CloseDevice(atadev
, name
);
1613 // Store drive size (for selective self-test only)
1614 state
.num_sectors
= get_num_sectors(&drive
);
1616 // Show if device in database, and use preset vendor attribute
1617 // options unless user has requested otherwise.
1618 if (cfg
.ignorepresets
)
1619 PrintOut(LOG_INFO
, "Device: %s, smartd database not searched (Directive: -P ignore).\n", name
);
1621 // do whatever applypresets decides to do.
1622 if (!apply_presets(&drive
, cfg
.attribute_defs
, cfg
.fix_firmwarebug
, fix_swapped_id
))
1623 PrintOut(LOG_INFO
, "Device: %s, not found in smartd database.\n", name
);
1625 PrintOut(LOG_INFO
, "Device: %s, found in smartd database.\n", name
);
1628 // Set default '-C 197[+]' if no '-C ID' is specified.
1629 if (!cfg
.curr_pending_set
)
1630 cfg
.curr_pending_id
= get_unc_attr_id(false, cfg
.attribute_defs
, cfg
.curr_pending_incr
);
1631 // Set default '-U 198[+]' if no '-U ID' is specified.
1632 if (!cfg
.offl_pending_set
)
1633 cfg
.offl_pending_id
= get_unc_attr_id(true, cfg
.attribute_defs
, cfg
.offl_pending_incr
);
1635 // If requested, show which presets would be used for this drive
1636 if (cfg
.showpresets
) {
1637 int savedebugmode
=debugmode
;
1638 PrintOut(LOG_INFO
, "Device %s: presets are:\n", name
);
1641 show_presets(&drive
, false);
1642 debugmode
=savedebugmode
;
1645 // see if drive supports SMART
1646 supported
=ataSmartSupport(&drive
);
1649 // drive does NOT support SMART
1650 PrintOut(LOG_INFO
,"Device: %s, lacks SMART capability\n",name
);
1652 // can't tell if drive supports SMART
1653 PrintOut(LOG_INFO
,"Device: %s, ATA IDENTIFY DEVICE words 82-83 don't specify if SMART capable.\n",name
);
1655 // should we proceed anyway?
1656 if (cfg
.permissive
) {
1657 PrintOut(LOG_INFO
,"Device: %s, proceeding since '-T permissive' Directive given.\n",name
);
1660 PrintOut(LOG_INFO
,"Device: %s, to proceed anyway, use '-T permissive' Directive.\n",name
);
1661 CloseDevice(atadev
, name
);
1666 if (ataEnableSmart(atadev
)) {
1667 // Enable SMART command has failed
1668 PrintOut(LOG_INFO
,"Device: %s, could not enable SMART capability\n",name
);
1669 CloseDevice(atadev
, name
);
1673 // disable device attribute autosave...
1674 if (cfg
.autosave
==1) {
1675 if (ataDisableAutoSave(atadev
))
1676 PrintOut(LOG_INFO
,"Device: %s, could not disable SMART Attribute Autosave.\n",name
);
1678 PrintOut(LOG_INFO
,"Device: %s, disabled SMART Attribute Autosave.\n",name
);
1681 // or enable device attribute autosave
1682 if (cfg
.autosave
==2) {
1683 if (ataEnableAutoSave(atadev
))
1684 PrintOut(LOG_INFO
,"Device: %s, could not enable SMART Attribute Autosave.\n",name
);
1686 PrintOut(LOG_INFO
,"Device: %s, enabled SMART Attribute Autosave.\n",name
);
1689 // capability check: SMART status
1690 if (cfg
.smartcheck
&& ataSmartStatus2(atadev
) == -1) {
1691 PrintOut(LOG_INFO
,"Device: %s, not capable of SMART Health Status check\n",name
);
1692 cfg
.smartcheck
= false;
1695 // capability check: Read smart values and thresholds. Note that
1696 // smart values are ALSO needed even if we ONLY want to know if the
1697 // device is self-test log or error-log capable! After ATA-5, this
1698 // information was ALSO reproduced in the IDENTIFY DEVICE response,
1699 // but sadly not for ATA-5. Sigh.
1701 // do we need to retain SMART data after returning from this routine?
1702 bool retainsmartdata
= (cfg
.usagefailed
|| cfg
.prefail
|| cfg
.usage
|| cfg
.tempdiff
|| cfg
.tempinfo
|| cfg
.tempcrit
);
1704 // do we need to get SMART data?
1705 bool smart_val_ok
= false;
1706 if ( retainsmartdata
|| cfg
.autoofflinetest
|| cfg
.selftest
|| cfg
.errorlog
1707 || cfg
.curr_pending_id
|| cfg
.offl_pending_id
) {
1709 if (ataReadSmartValues(atadev
, &state
.smartval
) ||
1710 ataReadSmartThresholds (atadev
, &state
.smartthres
)) {
1711 PrintOut(LOG_INFO
,"Device: %s, Read SMART Values and/or Thresholds Failed\n",name
);
1712 retainsmartdata
= cfg
.usagefailed
= cfg
.prefail
= cfg
.usage
= false;
1713 cfg
.tempdiff
= cfg
.tempinfo
= cfg
.tempcrit
= 0;
1714 cfg
.curr_pending_id
= cfg
.offl_pending_id
= 0;
1717 smart_val_ok
= true;
1719 // see if the necessary Attribute is there to monitor offline or
1720 // current pending sectors or temperature
1721 if (cfg
.curr_pending_id
&& ata_find_attr_index(cfg
.curr_pending_id
, state
.smartval
) < 0) {
1722 PrintOut(LOG_INFO
,"Device: %s, can't monitor Current Pending Sector count - no Attribute %d\n",
1723 name
, cfg
.curr_pending_id
);
1724 cfg
.curr_pending_id
= 0;
1727 if (cfg
.offl_pending_id
&& ata_find_attr_index(cfg
.offl_pending_id
, state
.smartval
) < 0) {
1728 PrintOut(LOG_INFO
,"Device: %s, can't monitor Offline Uncorrectable Sector count - no Attribute %d\n",
1729 name
, cfg
.offl_pending_id
);
1730 cfg
.offl_pending_id
= 0;
1733 if ( (cfg
.tempdiff
|| cfg
.tempinfo
|| cfg
.tempcrit
)
1734 && !ata_return_temperature_value(&state
.smartval
, cfg
.attribute_defs
)) {
1735 PrintOut(LOG_CRIT
, "Device: %s, can't monitor Temperature, ignoring -W Directive\n", name
);
1736 cfg
.tempdiff
= cfg
.tempinfo
= cfg
.tempcrit
= 0;
1740 // enable/disable automatic on-line testing
1741 if (cfg
.autoofflinetest
) {
1742 // is this an enable or disable request?
1743 const char *what
=(cfg
.autoofflinetest
==1)?"disable":"enable";
1745 PrintOut(LOG_INFO
,"Device: %s, could not %s SMART Automatic Offline Testing.\n",name
, what
);
1747 // if command appears unsupported, issue a warning...
1748 if (!isSupportAutomaticTimer(&state
.smartval
))
1749 PrintOut(LOG_INFO
,"Device: %s, SMART Automatic Offline Testing unsupported...\n",name
);
1750 // ... but then try anyway
1751 if ((cfg
.autoofflinetest
==1)?ataDisableAutoOffline(atadev
):ataEnableAutoOffline(atadev
))
1752 PrintOut(LOG_INFO
,"Device: %s, %s SMART Automatic Offline Testing failed.\n", name
, what
);
1754 PrintOut(LOG_INFO
,"Device: %s, %sd SMART Automatic Offline Testing.\n", name
, what
);
1758 // capability check: self-test-log
1762 // start with service disabled, and re-enable it if all works OK
1763 cfg
.selftest
= false;
1764 state
.selflogcount
= 0;
1765 state
.selfloghour
= 0;
1768 PrintOut(LOG_INFO
, "Device: %s, no SMART Self-Test log (SMART READ DATA failed); disabling -l selftest\n", name
);
1769 else if (!cfg
.permissive
&& !isSmartTestLogCapable(&state
.smartval
, &drive
))
1770 PrintOut(LOG_INFO
, "Device: %s, appears to lack SMART Self-Test log; disabling -l selftest (override with -T permissive Directive)\n", name
);
1771 else if ((retval
= SelfTestErrorCount(atadev
, name
, cfg
.fix_firmwarebug
)) < 0)
1772 PrintOut(LOG_INFO
, "Device: %s, no SMART Self-Test log; remove -l selftest Directive from smartd.conf\n", name
);
1774 cfg
.selftest
= true;
1775 state
.selflogcount
=SELFTEST_ERRORCOUNT(retval
);
1776 state
.selfloghour
=SELFTEST_ERRORHOURS(retval
);
1780 // capability check: ATA error log
1784 // start with service disabled, and re-enable it if all works OK
1785 cfg
.errorlog
= false;
1786 state
.ataerrorcount
=0;
1789 PrintOut(LOG_INFO
, "Device: %s, no SMART Error log (SMART READ DATA failed); disabling -l error\n", name
);
1790 else if (!cfg
.permissive
&& !isSmartErrorLogCapable(&state
.smartval
, &drive
))
1791 PrintOut(LOG_INFO
, "Device: %s, appears to lack SMART Error log; disabling -l error (override with -T permissive Directive)\n", name
);
1792 else if ((val
= ATAErrorCount(atadev
, name
, cfg
.fix_firmwarebug
)) < 0)
1793 PrintOut(LOG_INFO
, "Device: %s, no SMART Error log; remove -l error Directive from smartd.conf\n", name
);
1795 cfg
.errorlog
= true;
1796 state
.ataerrorcount
=val
;
1800 // capabilities check -- does it support powermode?
1801 if (cfg
.powermode
) {
1802 int powermode
= ataCheckPowerMode(atadev
);
1804 if (-1 == powermode
) {
1805 PrintOut(LOG_CRIT
, "Device: %s, no ATA CHECK POWER STATUS support, ignoring -n Directive\n", name
);
1808 else if (powermode
!=0 && powermode
!=0x80 && powermode
!=0xff) {
1809 PrintOut(LOG_CRIT
, "Device: %s, CHECK POWER STATUS returned %d, not ATA compliant, ignoring -n Directive\n",
1815 // If no tests available or selected, return
1816 if (!(cfg
.errorlog
|| cfg
.selftest
|| cfg
.smartcheck
||
1817 cfg
.usagefailed
|| cfg
.prefail
|| cfg
.usage
||
1818 cfg
.tempdiff
|| cfg
.tempinfo
|| cfg
.tempcrit
)) {
1819 CloseDevice(atadev
, name
);
1823 // tell user we are registering device
1824 PrintOut(LOG_INFO
,"Device: %s, is SMART capable. Adding to \"monitor\" list.\n",name
);
1826 // record number of device, type of device, increment device count
1827 if (cfg
.dev_type
.empty())
1828 cfg
.dev_type
= "ata";
1830 // close file descriptor
1831 CloseDevice(atadev
, name
);
1833 if (!state_path_prefix
.empty() || !attrlog_path_prefix
.empty()) {
1834 // Build file name for state file
1835 char model
[40+1], serial
[20+1];
1836 format_ata_string(model
, drive
.model
, sizeof(model
)-1, fix_swapped_id
);
1837 format_ata_string(serial
, drive
.serial_no
, sizeof(serial
)-1, fix_swapped_id
);
1838 std::replace_if(model
, model
+strlen(model
), not_allowed_in_filename
, '_');
1839 std::replace_if(serial
, serial
+strlen(serial
), not_allowed_in_filename
, '_');
1840 if (!state_path_prefix
.empty()) {
1841 cfg
.state_file
= strprintf("%s%s-%s.ata.state", state_path_prefix
.c_str(), model
, serial
);
1842 // Read previous state
1843 if (read_dev_state(cfg
.state_file
.c_str(), state
)) {
1844 PrintOut(LOG_INFO
, "Device: %s, state read from %s\n", name
, cfg
.state_file
.c_str());
1845 // Copy ATA attribute values to temp state
1846 state
.update_temp_state();
1849 if (!attrlog_path_prefix
.empty())
1850 cfg
.attrlog_file
= strprintf("%s%s-%s.ata.csv", attrlog_path_prefix
.c_str(), model
, serial
);
1853 // Start self-test regex check now if time was not read from state file
1854 if (!cfg
.test_regex
.empty() && !state
.scheduled_test_next_check
)
1855 state
.scheduled_test_next_check
= time(0);
1860 // on success, return 0. On failure, return >0. Never return <0,
1862 static int SCSIDeviceScan(dev_config
& cfg
, dev_state
& state
, scsi_device
* scsidev
)
1865 const char *device
= cfg
.name
.c_str();
1866 struct scsi_iec_mode_page iec
;
1869 // Device must be open
1871 // check that device is ready for commands. IE stores its stuff on
1873 if ((err
= scsiTestUnitReady(scsidev
))) {
1874 if (SIMPLE_ERR_NOT_READY
== err
)
1875 PrintOut(LOG_INFO
, "Device: %s, NOT READY (e.g. spun down); skip device\n", device
);
1876 else if (SIMPLE_ERR_NO_MEDIUM
== err
)
1877 PrintOut(LOG_INFO
, "Device: %s, NO MEDIUM present; skip device\n", device
);
1878 else if (SIMPLE_ERR_BECOMING_READY
== err
)
1879 PrintOut(LOG_INFO
, "Device: %s, BECOMING (but not yet) READY; skip device\n", device
);
1881 PrintOut(LOG_CRIT
, "Device: %s, failed Test Unit Ready [err=%d]\n", device
, err
);
1882 CloseDevice(scsidev
, device
);
1886 // Badly-conforming USB storage devices may fail this check.
1887 // The response to the following IE mode page fetch (current and
1888 // changeable values) is carefully examined. It has been found
1889 // that various USB devices that malform the response will lock up
1890 // if asked for a log page (e.g. temperature) so it is best to
1892 if (!(err
= scsiFetchIECmpage(scsidev
, &iec
, state
.modese_len
)))
1893 state
.modese_len
= iec
.modese_len
;
1894 else if (SIMPLE_ERR_BAD_FIELD
== err
)
1895 ; /* continue since it is reasonable not to support IE mpage */
1896 else { /* any other error (including malformed response) unreasonable */
1898 "Device: %s, Bad IEC (SMART) mode page, err=%d, skip device\n",
1900 CloseDevice(scsidev
, device
);
1904 // N.B. The following is passive (i.e. it doesn't attempt to turn on
1905 // smart if it is off). This may change to be the same as the ATA side.
1906 if (!scsi_IsExceptionControlEnabled(&iec
)) {
1907 PrintOut(LOG_INFO
, "Device: %s, IE (SMART) not enabled, skip device\n"
1908 "Try 'smartctl -s on %s' to turn on SMART features\n",
1910 CloseDevice(scsidev
, device
);
1914 // Flag that certain log pages are supported (information may be
1915 // available from other sources).
1916 if (0 == scsiLogSense(scsidev
, SUPPORTED_LPAGES
, 0, tBuf
, sizeof(tBuf
), 0)) {
1917 for (k
= 4; k
< tBuf
[3] + LOGPAGEHDRSIZE
; ++k
) {
1919 case TEMPERATURE_LPAGE
:
1920 state
.TempPageSupported
= 1;
1923 state
.SmartPageSupported
= 1;
1931 // record type of device
1932 if (cfg
.dev_type
.empty())
1933 cfg
.dev_type
= "scsi";
1935 // Check if scsiCheckIE() is going to work
1939 UINT8 currenttemp
= 0;
1942 if (scsiCheckIE(scsidev
, state
.SmartPageSupported
, state
.TempPageSupported
,
1943 &asc
, &ascq
, ¤ttemp
, &triptemp
)) {
1944 PrintOut(LOG_INFO
, "Device: %s, unexpectedly failed to read SMART values\n", device
);
1945 state
.SuppressReport
= 1;
1946 if (cfg
.tempdiff
|| cfg
.tempinfo
|| cfg
.tempcrit
) {
1947 PrintOut(LOG_CRIT
, "Device: %s, can't monitor Temperature, ignoring -W Directive\n", device
);
1948 cfg
.tempdiff
= cfg
.tempinfo
= cfg
.tempcrit
= 0;
1953 // capability check: self-test-log
1955 int retval
= scsiCountFailedSelfTests(scsidev
, 0);
1957 // no self-test log, turn off monitoring
1958 PrintOut(LOG_INFO
, "Device: %s, does not support SMART Self-Test Log.\n", device
);
1959 cfg
.selftest
= false;
1960 state
.selflogcount
= 0;
1961 state
.selfloghour
= 0;
1964 // register starting values to watch for changes
1965 state
.selflogcount
=SELFTEST_ERRORCOUNT(retval
);
1966 state
.selfloghour
=SELFTEST_ERRORHOURS(retval
);
1970 // disable autosave (set GLTSD bit)
1971 if (cfg
.autosave
==1){
1972 if (scsiSetControlGLTSD(scsidev
, 1, state
.modese_len
))
1973 PrintOut(LOG_INFO
,"Device: %s, could not disable autosave (set GLTSD bit).\n",device
);
1975 PrintOut(LOG_INFO
,"Device: %s, disabled autosave (set GLTSD bit).\n",device
);
1978 // or enable autosave (clear GLTSD bit)
1979 if (cfg
.autosave
==2){
1980 if (scsiSetControlGLTSD(scsidev
, 0, state
.modese_len
))
1981 PrintOut(LOG_INFO
,"Device: %s, could not enable autosave (clear GLTSD bit).\n",device
);
1983 PrintOut(LOG_INFO
,"Device: %s, enabled autosave (cleared GLTSD bit).\n",device
);
1986 // tell user we are registering device
1987 PrintOut(LOG_INFO
, "Device: %s, is SMART capable. Adding to \"monitor\" list.\n", device
);
1989 // TODO: Build file name for state file
1990 if (!state_path_prefix
.empty()) {
1991 PrintOut(LOG_INFO
, "Device: %s, persistence not yet supported for SCSI; ignoring -s option.\n", device
);
1993 // TODO: Build file name for attribute log file
1994 if (!attrlog_path_prefix
.empty()) {
1995 PrintOut(LOG_INFO
, "Device: %s, attribute log not yet supported for SCSI; ignoring -A option.\n", device
);
1998 // close file descriptor
1999 CloseDevice(scsidev
, device
);
2001 // Start self-test regex check now if time was not read from state file
2002 if (!cfg
.test_regex
.empty() && !state
.scheduled_test_next_check
)
2003 state
.scheduled_test_next_check
= time(0);
2008 // If the self-test log has got more self-test errors (or more recent
2009 // self-test errors) recorded, then notify user.
2010 static void CheckSelfTestLogs(const dev_config
& cfg
, dev_state
& state
, int newi
)
2012 const char * name
= cfg
.name
.c_str();
2016 MailWarning(cfg
, state
, 8, "Device: %s, Read SMART Self-Test Log Failed", name
);
2018 // old and new error counts
2019 int oldc
=state
.selflogcount
;
2020 int newc
=SELFTEST_ERRORCOUNT(newi
);
2022 // old and new error timestamps in hours
2023 int oldh
=state
.selfloghour
;
2024 int newh
=SELFTEST_ERRORHOURS(newi
);
2027 // increase in error count
2028 PrintOut(LOG_CRIT
, "Device: %s, Self-Test Log error count increased from %d to %d\n",
2030 MailWarning(cfg
, state
, 3, "Device: %s, Self-Test Log error count increased from %d to %d",
2032 state
.must_write
= true;
2033 } else if (oldh
!=newh
) {
2034 // more recent error
2035 // a 'more recent' error might actually be a smaller hour number,
2036 // if the hour number has wrapped.
2037 // There's still a bug here. You might just happen to run a new test
2038 // exactly 32768 hours after the previous failure, and have run exactly
2039 // 20 tests between the two, in which case smartd will miss the
2041 PrintOut(LOG_CRIT
, "Device: %s, new Self-Test Log error at hour timestamp %d\n",
2043 MailWarning(cfg
, state
, 3, "Device: %s, new Self-Test Log error at hour timestamp %d\n",
2045 state
.must_write
= true;
2048 // Needed since self-test error count may DECREASE. Hour might
2049 // also have changed.
2050 state
.selflogcount
= newc
;
2051 state
.selfloghour
= newh
;
2056 // Test types, ordered by priority.
2057 static const char test_type_chars
[] = "LncrSCO";
2058 const unsigned num_test_types
= sizeof(test_type_chars
)-1;
2060 // returns test type if time to do test of type testtype,
2061 // 0 if not time to do test.
2062 static char next_scheduled_test(const dev_config
& cfg
, dev_state
& state
, bool scsi
, time_t usetime
= 0)
2064 // check that self-testing has been requested
2065 if (cfg
.test_regex
.empty())
2068 // Exit if drive not capable of any test
2069 if ( state
.not_cap_long
&& state
.not_cap_short
&&
2070 (scsi
|| (state
.not_cap_conveyance
&& state
.not_cap_offline
)))
2073 // since we are about to call localtime(), be sure glibc is informed
2074 // of any timezone changes we make.
2076 FixGlibcTimeZoneBug();
2078 // Is it time for next check?
2079 time_t now
= (!usetime
? time(0) : usetime
);
2080 if (now
< state
.scheduled_test_next_check
)
2083 // Limit time check interval to 90 days
2084 if (state
.scheduled_test_next_check
+ (3600L*24*90) < now
)
2085 state
.scheduled_test_next_check
= now
- (3600L*24*90);
2087 // Check interval [state.scheduled_test_next_check, now] for scheduled tests
2089 time_t testtime
= 0; int testhour
= 0;
2090 int maxtest
= num_test_types
-1;
2092 for (time_t t
= state
.scheduled_test_next_check
; ; ) {
2093 struct tm
* tms
= localtime(&t
);
2094 // tm_wday is 0 (Sunday) to 6 (Saturday). We use 1 (Monday) to 7 (Sunday).
2095 int weekday
= (tms
->tm_wday
? tms
->tm_wday
: 7);
2096 for (int i
= 0; i
<= maxtest
; i
++) {
2097 // Skip if drive not capable of this test
2098 switch (test_type_chars
[i
]) {
2099 case 'L': if (state
.not_cap_long
) continue; break;
2100 case 'S': if (state
.not_cap_short
) continue; break;
2101 case 'C': if (scsi
|| state
.not_cap_conveyance
) continue; break;
2102 case 'O': if (scsi
|| state
.not_cap_offline
) continue; break;
2104 case 'r': if (scsi
|| state
.not_cap_selective
) continue; break;
2107 // Try match of "T/MM/DD/d/HH"
2109 snprintf(pattern
, sizeof(pattern
), "%c/%02d/%02d/%1d/%02d",
2110 test_type_chars
[i
], tms
->tm_mon
+1, tms
->tm_mday
, weekday
, tms
->tm_hour
);
2111 if (cfg
.test_regex
.full_match(pattern
)) {
2113 testtype
= pattern
[0];
2114 testtime
= t
; testhour
= tms
->tm_hour
;
2115 // Limit further matches to higher priority self-tests
2120 // Exit if no tests left or current time reached
2126 if ((t
+= 3600) > now
)
2130 // Do next check not before next hour.
2131 struct tm
* tmnow
= localtime(&now
);
2132 state
.scheduled_test_next_check
= now
+ (3600 - tmnow
->tm_min
*60 - tmnow
->tm_sec
);
2135 state
.must_write
= true;
2136 // Tell user if an old test was found.
2137 if (!usetime
&& !(testhour
== tmnow
->tm_hour
&& testtime
+ 3600 > now
)) {
2138 char datebuf
[DATEANDEPOCHLEN
]; dateandtimezoneepoch(datebuf
, testtime
);
2139 PrintOut(LOG_INFO
, "Device: %s, old test of type %c not run at %s, starting now.\n",
2140 cfg
.name
.c_str(), testtype
, datebuf
);
2147 // Print a list of future tests.
2148 static void PrintTestSchedule(const dev_config_vector
& configs
, dev_state_vector
& states
, const smart_device_list
& devices
)
2150 unsigned numdev
= configs
.size();
2153 std::vector
<int> testcnts(numdev
* num_test_types
, 0);
2155 PrintOut(LOG_INFO
, "\nNext scheduled self tests (at most 5 of each type per device):\n");
2157 // FixGlibcTimeZoneBug(); // done in PrintOut()
2158 time_t now
= time(0);
2159 char datenow
[DATEANDEPOCHLEN
], date
[DATEANDEPOCHLEN
];
2160 dateandtimezoneepoch(datenow
, now
);
2163 for (seconds
=checktime
; seconds
<3600L*24*90; seconds
+=checktime
) {
2164 // Check for each device whether a test will be run
2165 time_t testtime
= now
+ seconds
;
2166 for (unsigned i
= 0; i
< numdev
; i
++) {
2167 const dev_config
& cfg
= configs
.at(i
);
2168 dev_state
& state
= states
.at(i
);
2170 char testtype
= next_scheduled_test(cfg
, state
, devices
.at(i
)->is_scsi(), testtime
);
2171 if (testtype
&& (p
= strchr(test_type_chars
, testtype
))) {
2172 unsigned t
= (p
- test_type_chars
);
2173 // Report at most 5 tests of each type
2174 if (++testcnts
[i
*num_test_types
+ t
] <= 5) {
2175 dateandtimezoneepoch(date
, testtime
);
2176 PrintOut(LOG_INFO
, "Device: %s, will do test %d of type %c at %s\n", cfg
.name
.c_str(),
2177 testcnts
[i
*num_test_types
+ t
], testtype
, date
);
2184 dateandtimezoneepoch(date
, now
+seconds
);
2185 PrintOut(LOG_INFO
, "\nTotals [%s - %s]:\n", datenow
, date
);
2186 for (unsigned i
= 0; i
< numdev
; i
++) {
2187 const dev_config
& cfg
= configs
.at(i
);
2188 bool scsi
= devices
.at(i
)->is_scsi();
2189 for (unsigned t
= 0; t
< num_test_types
; t
++) {
2190 int cnt
= testcnts
[i
*num_test_types
+ t
];
2191 if (cnt
== 0 && !strchr((scsi
? "LS" : "LSCO"), test_type_chars
[t
]))
2193 PrintOut(LOG_INFO
, "Device: %s, will do %3d test%s of type %c\n", cfg
.name
.c_str(),
2194 cnt
, (cnt
==1?"":"s"), test_type_chars
[t
]);
2200 // Return zero on success, nonzero on failure. Perform offline (background)
2201 // short or long (extended) self test on given scsi device.
2202 static int DoSCSISelfTest(const dev_config
& cfg
, dev_state
& state
, scsi_device
* device
, char testtype
)
2205 const char *testname
= 0;
2206 const char *name
= cfg
.name
.c_str();
2209 if (scsiSelfTestInProgress(device
, &inProgress
)) {
2210 PrintOut(LOG_CRIT
, "Device: %s, does not support Self-Tests\n", name
);
2211 state
.not_cap_short
= state
.not_cap_long
= true;
2215 if (1 == inProgress
) {
2216 PrintOut(LOG_INFO
, "Device: %s, skip since Self-Test already in "
2217 "progress.\n", name
);
2223 testname
= "Short Self";
2224 retval
= scsiSmartShortSelfTest(device
);
2227 testname
= "Long Self";
2228 retval
= scsiSmartExtendSelfTest(device
);
2231 // If we can't do the test, exit
2232 if (NULL
== testname
) {
2233 PrintOut(LOG_CRIT
, "Device: %s, not capable of %c Self-Test\n", name
,
2238 if ((SIMPLE_ERR_BAD_OPCODE
== retval
) ||
2239 (SIMPLE_ERR_BAD_FIELD
== retval
)) {
2240 PrintOut(LOG_CRIT
, "Device: %s, not capable of %s-Test\n", name
,
2243 state
.not_cap_long
= true;
2245 state
.not_cap_short
= true;
2249 PrintOut(LOG_CRIT
, "Device: %s, execute %s-Test failed (err: %d)\n", name
,
2254 PrintOut(LOG_INFO
, "Device: %s, starting scheduled %s-Test.\n", name
, testname
);
2259 // Do an offline immediate or self-test. Return zero on success,
2260 // nonzero on failure.
2261 static int DoATASelfTest(const dev_config
& cfg
, dev_state
& state
, ata_device
* device
, char testtype
)
2263 const char *name
= cfg
.name
.c_str();
2265 // Read current smart data and check status/capability
2266 struct ata_smart_values data
;
2267 if (ataReadSmartValues(device
, &data
) || !(data
.offline_data_collection_capability
)) {
2268 PrintOut(LOG_CRIT
, "Device: %s, not capable of Offline or Self-Testing.\n", name
);
2272 // Check for capability to do the test
2273 int dotest
= -1, mode
= 0;
2274 const char *testname
= 0;
2277 testname
="Offline Immediate ";
2278 if (isSupportExecuteOfflineImmediate(&data
))
2279 dotest
=OFFLINE_FULL_SCAN
;
2281 state
.not_cap_offline
= true;
2284 testname
="Conveyance Self-";
2285 if (isSupportConveyanceSelfTest(&data
))
2286 dotest
=CONVEYANCE_SELF_TEST
;
2288 state
.not_cap_conveyance
= true;
2291 testname
="Short Self-";
2292 if (isSupportSelfTest(&data
))
2293 dotest
=SHORT_SELF_TEST
;
2295 state
.not_cap_short
= true;
2298 testname
="Long Self-";
2299 if (isSupportSelfTest(&data
))
2300 dotest
=EXTEND_SELF_TEST
;
2302 state
.not_cap_long
= true;
2305 case 'c': case 'n': case 'r':
2306 testname
= "Selective Self-";
2307 if (isSupportSelectiveSelfTest(&data
)) {
2308 dotest
= SELECTIVE_SELF_TEST
;
2310 case 'c': mode
= SEL_CONT
; break;
2311 case 'n': mode
= SEL_NEXT
; break;
2312 case 'r': mode
= SEL_REDO
; break;
2316 state
.not_cap_selective
= true;
2320 // If we can't do the test, exit
2322 PrintOut(LOG_CRIT
, "Device: %s, not capable of %sTest\n", name
, testname
);
2326 // If currently running a self-test, do not interrupt it to start another.
2327 if (15==(data
.self_test_exec_status
>> 4)) {
2328 if (cfg
.fix_firmwarebug
== FIX_SAMSUNG3
&& data
.self_test_exec_status
== 0xf0) {
2329 PrintOut(LOG_INFO
, "Device: %s, will not skip scheduled %sTest "
2330 "despite unclear Self-Test byte (SAMSUNG Firmware bug).\n", name
, testname
);
2332 PrintOut(LOG_INFO
, "Device: %s, skip scheduled %sTest; %1d0%% remaining of current Self-Test.\n",
2333 name
, testname
, (int)(data
.self_test_exec_status
& 0x0f));
2338 if (dotest
== SELECTIVE_SELF_TEST
) {
2340 ata_selective_selftest_args selargs
;
2341 selargs
.num_spans
= 1;
2342 selargs
.span
[0].mode
= mode
;
2343 if (ataWriteSelectiveSelfTestLog(device
, selargs
, &data
, state
.num_sectors
)) {
2344 PrintOut(LOG_CRIT
, "Device: %s, prepare %sTest failed\n", name
, testname
);
2347 uint64_t start
= selargs
.span
[0].start
, end
= selargs
.span
[0].end
;
2348 PrintOut(LOG_INFO
, "Device: %s, %s test span at LBA %"PRIu64
" - %"PRIu64
" (%"PRIu64
" sectors, %u%% - %u%% of disk).\n",
2349 name
, (selargs
.span
[0].mode
== SEL_NEXT
? "next" : "redo"),
2350 start
, end
, end
- start
+ 1,
2351 (unsigned)((100 * start
+ state
.num_sectors
/2) / state
.num_sectors
),
2352 (unsigned)((100 * end
+ state
.num_sectors
/2) / state
.num_sectors
));
2355 // execute the test, and return status
2356 int retval
= smartcommandhandler(device
, IMMEDIATE_OFFLINE
, dotest
, NULL
);
2358 PrintOut(LOG_CRIT
, "Device: %s, execute %sTest failed.\n", name
, testname
);
2362 if (testtype
!= 'O')
2363 // Log next self-test execution status
2364 state
.smartval
.self_test_exec_status
= 0xff;
2366 PrintOut(LOG_INFO
, "Device: %s, starting scheduled %sTest.\n", name
, testname
);
2370 // Check pending sector count attribute values (-C, -U directives).
2371 static void check_pending(const dev_config
& cfg
, dev_state
& state
,
2372 unsigned char id
, bool increase_only
,
2373 const ata_smart_values
& smartval
,
2374 int mailtype
, const char * msg
)
2376 // Find attribute index
2377 int i
= ata_find_attr_index(id
, smartval
);
2378 if (!(i
>= 0 && ata_find_attr_index(id
, state
.smartval
) == i
))
2381 // No report if no sectors pending.
2382 uint64_t rawval
= ata_get_attr_raw_value(smartval
.vendor_attributes
[i
], cfg
.attribute_defs
);
2386 // If attribute is not reset, report only sector count increases.
2387 uint64_t prev_rawval
= ata_get_attr_raw_value(state
.smartval
.vendor_attributes
[i
], cfg
.attribute_defs
);
2388 if (!(!increase_only
|| prev_rawval
< rawval
))
2392 std::string s
= strprintf("Device: %s, %"PRId64
" %s", cfg
.name
.c_str(), rawval
, msg
);
2393 if (prev_rawval
> 0 && rawval
!= prev_rawval
)
2394 s
+= strprintf(" (changed %+"PRId64
")", rawval
- prev_rawval
);
2396 PrintOut(LOG_CRIT
, "%s\n", s
.c_str());
2397 MailWarning(cfg
, state
, mailtype
, "%s\n", s
.c_str());
2398 state
.must_write
= true;
2401 // Format Temperature value
2402 static const char * fmt_temp(unsigned char x
, char * buf
)
2407 sprintf(buf
, "%u", x
);
2411 // Check Temperature limits
2412 static void CheckTemperature(const dev_config
& cfg
, dev_state
& state
, unsigned char currtemp
, unsigned char triptemp
)
2414 if (!(0 < currtemp
&& currtemp
< 255)) {
2415 PrintOut(LOG_INFO
, "Device: %s, failed to read Temperature\n", cfg
.name
.c_str());
2419 // Update Max Temperature
2420 const char * minchg
= "", * maxchg
= "";
2421 if (currtemp
> state
.tempmax
) {
2424 state
.tempmax
= currtemp
;
2425 state
.must_write
= true;
2429 if (!state
.temperature
) {
2431 if (!state
.tempmin
|| currtemp
< state
.tempmin
)
2432 // Delay Min Temperature update by ~ 30 minutes.
2433 state
.tempmin_delay
= time(0) + CHECKTIME
- 60;
2434 PrintOut(LOG_INFO
, "Device: %s, initial Temperature is %d Celsius (Min/Max %s/%u%s)\n",
2435 cfg
.name
.c_str(), (int)currtemp
, fmt_temp(state
.tempmin
, buf
), state
.tempmax
, maxchg
);
2437 PrintOut(LOG_INFO
, " [trip Temperature is %d Celsius]\n", (int)triptemp
);
2438 state
.temperature
= currtemp
;
2441 if (state
.tempmin_delay
) {
2442 // End Min Temperature update delay if ...
2443 if ( (state
.tempmin
&& currtemp
> state
.tempmin
) // current temp exceeds recorded min,
2444 || (state
.tempmin_delay
<= time(0))) { // or delay time is over.
2445 state
.tempmin_delay
= 0;
2447 state
.tempmin
= 255;
2451 // Update Min Temperature
2452 if (!state
.tempmin_delay
&& currtemp
< state
.tempmin
) {
2453 state
.tempmin
= currtemp
;
2454 state
.must_write
= true;
2455 if (currtemp
!= state
.temperature
)
2460 if (cfg
.tempdiff
&& (*minchg
|| *maxchg
|| abs((int)currtemp
- (int)state
.temperature
) >= cfg
.tempdiff
)) {
2461 PrintOut(LOG_INFO
, "Device: %s, Temperature changed %+d Celsius to %u Celsius (Min/Max %s%s/%u%s)\n",
2462 cfg
.name
.c_str(), (int)currtemp
-(int)state
.temperature
, currtemp
, fmt_temp(state
.tempmin
, buf
), minchg
, state
.tempmax
, maxchg
);
2463 state
.temperature
= currtemp
;
2468 if (cfg
.tempcrit
&& currtemp
>= cfg
.tempcrit
) {
2469 PrintOut(LOG_CRIT
, "Device: %s, Temperature %u Celsius reached critical limit of %u Celsius (Min/Max %s%s/%u%s)\n",
2470 cfg
.name
.c_str(), currtemp
, cfg
.tempcrit
, fmt_temp(state
.tempmin
, buf
), minchg
, state
.tempmax
, maxchg
);
2471 MailWarning(cfg
, state
, 12, "Device: %s, Temperature %d Celsius reached critical limit of %u Celsius (Min/Max %s%s/%u%s)\n",
2472 cfg
.name
.c_str(), currtemp
, cfg
.tempcrit
, fmt_temp(state
.tempmin
, buf
), minchg
, state
.tempmax
, maxchg
);
2474 else if (cfg
.tempinfo
&& currtemp
>= cfg
.tempinfo
) {
2475 PrintOut(LOG_INFO
, "Device: %s, Temperature %u Celsius reached limit of %u Celsius (Min/Max %s%s/%u%s)\n",
2476 cfg
.name
.c_str(), currtemp
, cfg
.tempinfo
, fmt_temp(state
.tempmin
, buf
), minchg
, state
.tempmax
, maxchg
);
2480 // Check normalized and raw attribute values.
2481 static void check_attribute(const dev_config
& cfg
, dev_state
& state
,
2482 const ata_smart_attribute
& attr
,
2483 const ata_smart_attribute
& prev
,
2484 const ata_smart_threshold_entry
& thre
)
2486 // Check attribute and threshold
2487 ata_attr_state attrstate
= ata_get_attr_state(attr
, thre
, cfg
.attribute_defs
);
2488 if (attrstate
== ATTRSTATE_NON_EXISTING
)
2491 // If requested, check for usage attributes that have failed.
2492 if ( cfg
.usagefailed
&& attrstate
== ATTRSTATE_FAILED_NOW
2493 && !cfg
.monitor_attr_flags
.is_set(attr
.id
, MONITOR_IGN_FAILUSE
)) {
2494 std::string attrname
= ata_get_smart_attr_name(attr
.id
, cfg
.attribute_defs
);
2495 PrintOut(LOG_CRIT
, "Device: %s, Failed SMART usage Attribute: %d %s.\n", cfg
.name
.c_str(), attr
.id
, attrname
.c_str());
2496 MailWarning(cfg
, state
, 2, "Device: %s, Failed SMART usage Attribute: %d %s.", cfg
.name
.c_str(), attr
.id
, attrname
.c_str());
2497 state
.must_write
= true;
2500 // Return if we're not tracking this type of attribute
2501 bool prefail
= !!ATTRIBUTE_FLAGS_PREFAILURE(attr
.flags
);
2502 if (!( ( prefail
&& cfg
.prefail
)
2503 || (!prefail
&& cfg
.usage
)))
2506 // Return if '-I ID' was specified
2507 if (cfg
.monitor_attr_flags
.is_set(attr
.id
, MONITOR_IGNORE
))
2510 // Issue warning if they don't have the same ID in all structures.
2511 if (attr
.id
!= prev
.id
|| attrstate
== ATTRSTATE_BAD_THRESHOLD
) {
2512 PrintOut(LOG_INFO
,"Device: %s, same Attribute has different ID numbers: %d = %d = %d\n",
2513 cfg
.name
.c_str(), attr
.id
, prev
.id
, thre
.id
);
2517 // Compare normalized values if valid.
2518 bool valchanged
= false;
2519 if (attrstate
> ATTRSTATE_NO_NORMVAL
) {
2520 if (attr
.current
!= prev
.current
)
2524 // Compare raw values if requested.
2525 bool rawchanged
= false;
2526 if (cfg
.monitor_attr_flags
.is_set(attr
.id
, MONITOR_RAW
)) {
2527 if ( ata_get_attr_raw_value(attr
, cfg
.attribute_defs
)
2528 != ata_get_attr_raw_value(prev
, cfg
.attribute_defs
))
2532 // Return if no change
2533 if (!(valchanged
|| rawchanged
))
2536 // Format value strings
2537 std::string currstr
, prevstr
;
2538 if (attrstate
== ATTRSTATE_NO_NORMVAL
) {
2539 // Print raw values only
2540 currstr
= strprintf("%s (Raw)",
2541 ata_format_attr_raw_value(attr
, cfg
.attribute_defs
).c_str());
2542 prevstr
= strprintf("%s (Raw)",
2543 ata_format_attr_raw_value(prev
, cfg
.attribute_defs
).c_str());
2545 else if (cfg
.monitor_attr_flags
.is_set(attr
.id
, MONITOR_RAW_PRINT
)) {
2546 // Print normalized and raw values
2547 currstr
= strprintf("%d [Raw %s]", attr
.current
,
2548 ata_format_attr_raw_value(attr
, cfg
.attribute_defs
).c_str());
2549 prevstr
= strprintf("%d [Raw %s]", prev
.current
,
2550 ata_format_attr_raw_value(prev
, cfg
.attribute_defs
).c_str());
2553 // Print normalized values only
2554 currstr
= strprintf("%d", attr
.current
);
2555 prevstr
= strprintf("%d", prev
.current
);
2559 std::string msg
= strprintf("Device: %s, SMART %s Attribute: %d %s changed from %s to %s",
2560 cfg
.name
.c_str(), (prefail
? "Prefailure" : "Usage"), attr
.id
,
2561 ata_get_smart_attr_name(attr
.id
, cfg
.attribute_defs
).c_str(),
2562 prevstr
.c_str(), currstr
.c_str());
2564 // Report this change as critical ?
2565 if ( (valchanged
&& cfg
.monitor_attr_flags
.is_set(attr
.id
, MONITOR_AS_CRIT
))
2566 || (rawchanged
&& cfg
.monitor_attr_flags
.is_set(attr
.id
, MONITOR_RAW_AS_CRIT
))) {
2567 PrintOut(LOG_CRIT
, "%s\n", msg
.c_str());
2568 MailWarning(cfg
, state
, 2, "%s", msg
.c_str());
2571 PrintOut(LOG_INFO
, "%s\n", msg
.c_str());
2573 state
.must_write
= true;
2577 static int ATACheckDevice(const dev_config
& cfg
, dev_state
& state
, ata_device
* atadev
, bool allow_selftests
)
2579 const char * name
= cfg
.name
.c_str();
2581 // If user has asked, test the email warning system
2583 MailWarning(cfg
, state
, 0, "TEST EMAIL from smartd for device: %s", name
);
2585 // if we can't open device, fail gracefully rather than hard --
2586 // perhaps the next time around we'll be able to open it. ATAPI
2587 // cd/dvd devices will hang awaiting media if O_NONBLOCK is not
2588 // given (see linux cdrom driver).
2589 if (!atadev
->open()) {
2590 PrintOut(LOG_INFO
, "Device: %s, open() failed: %s\n", name
, atadev
->get_errmsg());
2591 MailWarning(cfg
, state
, 9, "Device: %s, unable to open device", name
);
2593 } else if (debugmode
)
2594 PrintOut(LOG_INFO
,"Device: %s, opened ATA device\n", name
);
2596 // user may have requested (with the -n Directive) to leave the disk
2597 // alone if it is in idle or sleeping mode. In this case check the
2598 // power mode and exit without check if needed
2599 if (cfg
.powermode
&& !state
.powermodefail
) {
2600 int dontcheck
=0, powermode
=ataCheckPowerMode(atadev
);
2601 const char * mode
= 0;
2602 if (0 <= powermode
&& powermode
< 0xff) {
2603 // wait for possible spin up and check again
2606 powermode2
= ataCheckPowerMode(atadev
);
2607 if (powermode2
> powermode
)
2608 PrintOut(LOG_INFO
, "Device: %s, CHECK POWER STATUS spins up disk (0x%02x -> 0x%02x)\n", name
, powermode
, powermode2
);
2609 powermode
= powermode2
;
2616 if (cfg
.powermode
>=1)
2622 if (cfg
.powermode
>=2)
2628 if (cfg
.powermode
>=3)
2633 mode
="ACTIVE or IDLE";
2637 PrintOut(LOG_CRIT
, "Device: %s, CHECK POWER STATUS returned %d, not ATA compliant, ignoring -n Directive\n",
2639 state
.powermodefail
= true;
2643 // if we are going to skip a check, return now
2645 // skip at most powerskipmax checks
2646 if (!cfg
.powerskipmax
|| state
.powerskipcnt
<cfg
.powerskipmax
) {
2647 CloseDevice(atadev
, name
);
2648 if (!state
.powerskipcnt
&& !cfg
.powerquiet
) // report first only and avoid waking up system disk
2649 PrintOut(LOG_INFO
, "Device: %s, is in %s mode, suspending checks\n", name
, mode
);
2650 state
.powerskipcnt
++;
2654 PrintOut(LOG_INFO
, "Device: %s, %s mode ignored due to reached limit of skipped checks (%d check%s skipped)\n",
2655 name
, mode
, state
.powerskipcnt
, (state
.powerskipcnt
==1?"":"s"));
2657 state
.powerskipcnt
= 0;
2658 state
.tempmin_delay
= time(0) + CHECKTIME
- 60; // Delay Min Temperature update
2660 else if (state
.powerskipcnt
) {
2661 PrintOut(LOG_INFO
, "Device: %s, is back in %s mode, resuming checks (%d check%s skipped)\n",
2662 name
, mode
, state
.powerskipcnt
, (state
.powerskipcnt
==1?"":"s"));
2663 state
.powerskipcnt
= 0;
2664 state
.tempmin_delay
= time(0) + CHECKTIME
- 60; // Delay Min Temperature update
2668 // check smart status
2669 if (cfg
.smartcheck
) {
2670 int status
=ataSmartStatus2(atadev
);
2672 PrintOut(LOG_INFO
,"Device: %s, not capable of SMART self-check\n",name
);
2673 MailWarning(cfg
, state
, 5, "Device: %s, not capable of SMART self-check", name
);
2674 state
.must_write
= true;
2676 else if (status
==1){
2677 PrintOut(LOG_CRIT
, "Device: %s, FAILED SMART self-check. BACK UP DATA NOW!\n", name
);
2678 MailWarning(cfg
, state
, 1, "Device: %s, FAILED SMART self-check. BACK UP DATA NOW!", name
);
2679 state
.must_write
= true;
2683 // Check everything that depends upon SMART Data (eg, Attribute values)
2684 if ( cfg
.usagefailed
|| cfg
.prefail
|| cfg
.usage
2685 || cfg
.curr_pending_id
|| cfg
.offl_pending_id
2686 || cfg
.tempdiff
|| cfg
.tempinfo
|| cfg
.tempcrit
|| cfg
.selftest
) {
2688 // Read current attribute values.
2689 ata_smart_values curval
;
2690 if (ataReadSmartValues(atadev
, &curval
)){
2691 PrintOut(LOG_CRIT
, "Device: %s, failed to read SMART Attribute Data\n", name
);
2692 MailWarning(cfg
, state
, 6, "Device: %s, failed to read SMART Attribute Data", name
);
2693 state
.must_write
= true;
2696 // look for current or offline pending sectors
2697 if (cfg
.curr_pending_id
)
2698 check_pending(cfg
, state
, cfg
.curr_pending_id
, cfg
.curr_pending_incr
, curval
, 10,
2699 (!cfg
.curr_pending_incr
? "Currently unreadable (pending) sectors"
2700 : "Total unreadable (pending) sectors" ));
2702 if (cfg
.offl_pending_id
)
2703 check_pending(cfg
, state
, cfg
.offl_pending_id
, cfg
.offl_pending_incr
, curval
, 11,
2704 (!cfg
.offl_pending_incr
? "Offline uncorrectable sectors"
2705 : "Total offline uncorrectable sectors"));
2707 // check temperature limits
2708 if (cfg
.tempdiff
|| cfg
.tempinfo
|| cfg
.tempcrit
)
2709 CheckTemperature(cfg
, state
, ata_return_temperature_value(&curval
, cfg
.attribute_defs
), 0);
2711 if (cfg
.usagefailed
|| cfg
.prefail
|| cfg
.usage
) {
2713 // look for failed usage attributes, or track usage or prefail attributes
2714 for (int i
= 0; i
< NUMBER_ATA_SMART_ATTRIBUTES
; i
++) {
2715 check_attribute(cfg
, state
,
2716 curval
.vendor_attributes
[i
],
2717 state
.smartval
.vendor_attributes
[i
],
2718 state
.smartthres
.thres_entries
[i
]);
2722 // Log changes of self-test execution status
2723 if ( curval
.self_test_exec_status
!= state
.smartval
.self_test_exec_status
2724 || (!allow_selftests
&& curval
.self_test_exec_status
!= 0x00) )
2725 log_self_test_exec_status(name
, curval
.self_test_exec_status
);
2728 // Save the new values into *drive for the next time around
2729 state
.smartval
= curval
;
2734 // check if number of selftest errors has increased (note: may also DECREASE)
2736 CheckSelfTestLogs(cfg
, state
, SelfTestErrorCount(atadev
, name
, cfg
.fix_firmwarebug
));
2738 // check if number of ATA errors has increased
2741 int newc
, oldc
= state
.ataerrorcount
;
2743 // new number of errors
2744 newc
= ATAErrorCount(atadev
, name
, cfg
.fix_firmwarebug
);
2746 // did command fail?
2748 // lack of PrintOut here is INTENTIONAL
2749 MailWarning(cfg
, state
, 7, "Device: %s, Read SMART Error Log Failed", name
);
2751 // has error count increased?
2753 PrintOut(LOG_CRIT
, "Device: %s, ATA error count increased from %d to %d\n",
2755 MailWarning(cfg
, state
, 4, "Device: %s, ATA error count increased from %d to %d",
2757 state
.must_write
= true;
2760 // this last line is probably not needed, count always increases
2762 state
.ataerrorcount
=newc
;
2765 // if the user has asked, and device is capable (or we're not yet
2766 // sure) check whether a self test should be done now.
2767 if (allow_selftests
&& !cfg
.test_regex
.empty()) {
2768 char testtype
= next_scheduled_test(cfg
, state
, false/*!scsi*/);
2770 DoATASelfTest(cfg
, state
, atadev
, testtype
);
2773 // Don't leave device open -- the OS/user may want to access it
2774 // before the next smartd cycle!
2775 CloseDevice(atadev
, name
);
2777 // Copy ATA attribute values to persistent state
2778 state
.update_persistent_state();
2783 static int SCSICheckDevice(const dev_config
& cfg
, dev_state
& state
, scsi_device
* scsidev
, bool allow_selftests
)
2788 const char * name
= cfg
.name
.c_str();
2791 // If the user has asked for it, test the email warning system
2793 MailWarning(cfg
, state
, 0, "TEST EMAIL from smartd for device: %s", name
);
2795 // if we can't open device, fail gracefully rather than hard --
2796 // perhaps the next time around we'll be able to open it
2797 if (!scsidev
->open()) {
2798 PrintOut(LOG_INFO
, "Device: %s, open() failed: %s\n", name
, scsidev
->get_errmsg());
2799 MailWarning(cfg
, state
, 9, "Device: %s, unable to open device", name
);
2801 } else if (debugmode
)
2802 PrintOut(LOG_INFO
,"Device: %s, opened SCSI device\n", name
);
2806 if (!state
.SuppressReport
) {
2807 if (scsiCheckIE(scsidev
, state
.SmartPageSupported
, state
.TempPageSupported
,
2808 &asc
, &ascq
, ¤ttemp
, &triptemp
)) {
2809 PrintOut(LOG_INFO
, "Device: %s, failed to read SMART values\n",
2811 MailWarning(cfg
, state
, 6, "Device: %s, failed to read SMART values", name
);
2812 state
.SuppressReport
= 1;
2816 cp
= scsiGetIEString(asc
, ascq
);
2818 PrintOut(LOG_CRIT
, "Device: %s, SMART Failure: %s\n", name
, cp
);
2819 MailWarning(cfg
, state
, 1,"Device: %s, SMART Failure: %s", name
, cp
);
2820 } else if (debugmode
)
2821 PrintOut(LOG_INFO
,"Device: %s, non-SMART asc,ascq: %d,%d\n",
2822 name
, (int)asc
, (int)ascq
);
2823 } else if (debugmode
)
2824 PrintOut(LOG_INFO
,"Device: %s, SMART health: passed\n", name
);
2826 // check temperature limits
2827 if (cfg
.tempdiff
|| cfg
.tempinfo
|| cfg
.tempcrit
)
2828 CheckTemperature(cfg
, state
, currenttemp
, triptemp
);
2830 // check if number of selftest errors has increased (note: may also DECREASE)
2832 CheckSelfTestLogs(cfg
, state
, scsiCountFailedSelfTests(scsidev
, 0));
2834 if (allow_selftests
&& !cfg
.test_regex
.empty()) {
2835 char testtype
= next_scheduled_test(cfg
, state
, true/*scsi*/);
2837 DoSCSISelfTest(cfg
, state
, scsidev
, testtype
);
2839 CloseDevice(scsidev
, name
);
2843 // Checks the SMART status of all ATA and SCSI devices
2844 static void CheckDevicesOnce(const dev_config_vector
& configs
, dev_state_vector
& states
,
2845 smart_device_list
& devices
, bool allow_selftests
)
2847 for (unsigned i
= 0; i
< configs
.size(); i
++) {
2848 const dev_config
& cfg
= configs
.at(i
);
2849 dev_state
& state
= states
.at(i
);
2850 smart_device
* dev
= devices
.at(i
);
2852 ATACheckDevice(cfg
, state
, dev
->to_ata(), allow_selftests
);
2853 else if (dev
->is_scsi())
2854 SCSICheckDevice(cfg
, state
, dev
->to_scsi(), allow_selftests
);
2858 // Set if Initialize() was called
2859 static bool is_initialized
= false;
2861 // Does initialization right after fork to daemon mode
2862 void Initialize(time_t *wakeuptime
){
2864 // Call Goodbye() on exit
2865 is_initialized
= true;
2871 // install signal handlers. On Solaris, can't use signal() because
2872 // it resets the handler to SIG_DFL after each call. So use sigset()
2873 // instead. So SIGNALFN()==signal() or SIGNALFN()==sigset().
2875 // normal and abnormal exit
2876 if (SIGNALFN(SIGTERM
, sighandler
)==SIG_IGN
)
2877 SIGNALFN(SIGTERM
, SIG_IGN
);
2878 if (SIGNALFN(SIGQUIT
, sighandler
)==SIG_IGN
)
2879 SIGNALFN(SIGQUIT
, SIG_IGN
);
2881 // in debug mode, <CONTROL-C> ==> HUP
2882 if (SIGNALFN(SIGINT
, debugmode
?HUPhandler
:sighandler
)==SIG_IGN
)
2883 SIGNALFN(SIGINT
, SIG_IGN
);
2885 // Catch HUP and USR1
2886 if (SIGNALFN(SIGHUP
, HUPhandler
)==SIG_IGN
)
2887 SIGNALFN(SIGHUP
, SIG_IGN
);
2888 if (SIGNALFN(SIGUSR1
, USR1handler
)==SIG_IGN
)
2889 SIGNALFN(SIGUSR1
, SIG_IGN
);
2891 if (SIGNALFN(SIGUSR2
, USR2handler
)==SIG_IGN
)
2892 SIGNALFN(SIGUSR2
, SIG_IGN
);
2895 // initialize wakeup time to CURRENT time
2896 *wakeuptime
=time(NULL
);
2902 // Toggle debug mode implemented for native windows only
2903 // (there is no easy way to reopen tty on *nix)
2904 static void ToggleDebugMode()
2907 PrintOut(LOG_INFO
,"Signal USR2 - enabling debug mode\n");
2908 if (!daemon_enable_console("smartd [Debug]")) {
2910 daemon_signal(SIGINT
, HUPhandler
);
2911 PrintOut(LOG_INFO
,"smartd debug mode enabled, PID=%d\n", getpid());
2914 PrintOut(LOG_INFO
,"enable console failed\n");
2916 else if (debugmode
== 1) {
2917 daemon_disable_console();
2919 daemon_signal(SIGINT
, sighandler
);
2920 PrintOut(LOG_INFO
,"Signal USR2 - debug mode disabled\n");
2923 PrintOut(LOG_INFO
,"Signal USR2 - debug mode %d not changed\n", debugmode
);
2927 static time_t dosleep(time_t wakeuptime
, bool & sigwakeup
)
2929 // If past wake-up-time, compute next wake-up-time
2930 time_t timenow
=time(NULL
);
2931 while (wakeuptime
<=timenow
){
2932 int intervals
=1+(timenow
-wakeuptime
)/checktime
;
2933 wakeuptime
+=intervals
*checktime
;
2936 // sleep until we catch SIGUSR1 or have completed sleeping
2937 while (timenow
<wakeuptime
&& !caughtsigUSR1
&& !caughtsigHUP
&& !caughtsigEXIT
){
2939 // protect user again system clock being adjusted backwards
2940 if (wakeuptime
>timenow
+checktime
){
2941 PrintOut(LOG_CRIT
, "System clock time adjusted to the past. Resetting next wakeup time.\n");
2942 wakeuptime
=timenow
+checktime
;
2945 // Exit sleep when time interval has expired or a signal is received
2946 sleep(wakeuptime
-timenow
);
2949 // toggle debug mode?
2950 if (caughtsigUSR2
) {
2959 // if we caught a SIGUSR1 then print message and clear signal
2961 PrintOut(LOG_INFO
,"Signal USR1 - checking devices now rather than in %d seconds.\n",
2962 wakeuptime
-timenow
>0?(int)(wakeuptime
-timenow
):0);
2967 // return adjusted wakeuptime
2971 // Print out a list of valid arguments for the Directive d
2972 void printoutvaliddirectiveargs(int priority
, char d
) {
2976 PrintOut(priority
, "never[,N][,q], sleep[,N][,q], standby[,N][,q], idle[,N][,q]");
2979 PrintOut(priority
, "valid_regular_expression");
2982 PrintOut(priority
, "%s", smi()->get_valid_dev_types_str().c_str());
2985 PrintOut(priority
, "normal, permissive");
2989 PrintOut(priority
, "on, off");
2992 PrintOut(priority
, "error, selftest");
2995 PrintOut(priority
, "\"once\", \"daily\", \"diminishing\", \"test\", \"exec\"");
2998 PrintOut(priority
, "\n%s\n", create_vendor_attribute_arg_list().c_str());
3001 PrintOut(priority
, "use, ignore, show, showall");
3004 PrintOut(priority
, "none, samsung, samsung2, samsung3");
3009 // exits with an error message, or returns integer value of token
3010 int GetInteger(const char *arg
, const char *name
, const char *token
, int lineno
, const char *configfile
,
3011 int min
, int max
, char * suffix
= 0)
3013 // make sure argument is there
3015 PrintOut(LOG_CRIT
,"File %s line %d (drive %s): Directive: %s takes integer argument from %d to %d.\n",
3016 configfile
, lineno
, name
, token
, min
, max
);
3020 // get argument value (base 10), check that it's integer, and in-range
3022 int val
= strtol(arg
,&endptr
,10);
3024 // optional suffix present?
3026 if (!strcmp(endptr
, suffix
))
3027 endptr
+= strlen(suffix
);
3032 if (!(!*endptr
&& min
<= val
&& val
<= max
)) {
3033 PrintOut(LOG_CRIT
,"File %s line %d (drive %s): Directive: %s has argument: %s; needs integer from %d to %d.\n",
3034 configfile
, lineno
, name
, token
, arg
, min
, max
);
3038 // all is well; return value
3043 // Get 1-3 small integer(s) for '-W' directive
3044 int Get3Integers(const char *arg
, const char *name
, const char *token
, int lineno
, const char *configfile
,
3045 unsigned char * val1
, unsigned char * val2
, unsigned char * val3
){
3046 unsigned v1
= 0, v2
= 0, v3
= 0;
3047 int n1
= -1, n2
= -1, n3
= -1, len
;
3049 PrintOut(LOG_CRIT
,"File %s line %d (drive %s): Directive: %s takes 1-3 integer argument(s) from 0 to 255.\n",
3050 configfile
, lineno
, name
, token
);
3055 if (!( sscanf(arg
, "%u%n,%u%n,%u%n", &v1
, &n1
, &v2
, &n2
, &v3
, &n3
) >= 1
3056 && (n1
== len
|| n2
== len
|| n3
== len
) && v1
<= 255 && v2
<= 255 && v3
<= 255)) {
3057 PrintOut(LOG_CRIT
,"File %s line %d (drive %s): Directive: %s has argument: %s; needs 1-3 integer(s) from 0 to 255.\n",
3058 configfile
, lineno
, name
, token
, arg
);
3061 *val1
= (unsigned char)v1
; *val2
= (unsigned char)v2
; *val3
= (unsigned char)v3
;
3066 // This function returns 1 if it has correctly parsed one token (and
3067 // any arguments), else zero if no tokens remain. It returns -1 if an
3068 // error was encountered.
3069 static int ParseToken(char * token
, dev_config
& cfg
)
3072 const char * name
= cfg
.name
.c_str();
3073 int lineno
=cfg
.lineno
;
3074 const char *delim
= " \n\t";
3077 const char *arg
= 0;
3079 // is the rest of the line a comment
3083 // is the token not recognized?
3084 if (*token
!='-' || strlen(token
)!=2) {
3085 PrintOut(LOG_CRIT
,"File %s line %d (drive %s): unknown Directive: %s\n",
3086 configfile
, lineno
, name
, token
);
3087 PrintOut(LOG_CRIT
, "Run smartd -D to print a list of valid Directives.\n");
3091 // token we will be parsing:
3094 // parse the token and swallow its argument
3096 char plus
[] = "+", excl
[] = "!";
3100 // monitor current pending sector count (default 197)
3101 if ((val
= GetInteger(arg
=strtok(NULL
,delim
), name
, token
, lineno
, configfile
, 0, 255, plus
)) < 0)
3103 cfg
.curr_pending_id
= (unsigned char)val
;
3104 cfg
.curr_pending_incr
= (*plus
== '+');
3105 cfg
.curr_pending_set
= true;
3108 // monitor offline uncorrectable sectors (default 198)
3109 if ((val
= GetInteger(arg
=strtok(NULL
,delim
), name
, token
, lineno
, configfile
, 0, 255, plus
)) < 0)
3111 cfg
.offl_pending_id
= (unsigned char)val
;
3112 cfg
.offl_pending_incr
= (*plus
== '+');
3113 cfg
.offl_pending_set
= true;
3116 // Set tolerance level for SMART command failures
3117 if ((arg
= strtok(NULL
, delim
)) == NULL
) {
3119 } else if (!strcmp(arg
, "normal")) {
3120 // Normal mode: exit on failure of a mandatory S.M.A.R.T. command, but
3121 // not on failure of an optional S.M.A.R.T. command.
3122 // This is the default so we don't need to actually do anything here.
3123 cfg
.permissive
= false;
3124 } else if (!strcmp(arg
, "permissive")) {
3125 // Permissive mode; ignore errors from Mandatory SMART commands
3126 cfg
.permissive
= true;
3132 // specify the device type
3133 if ((arg
= strtok(NULL
, delim
)) == NULL
) {
3135 } else if (!strcmp(arg
, "removable")) {
3136 cfg
.removable
= true;
3143 if ((arg
= strtok(NULL
, delim
)) == NULL
) {
3145 } else if (!strcmp(arg
, "none")) {
3146 cfg
.fix_firmwarebug
= FIX_NONE
;
3147 } else if (!strcmp(arg
, "samsung")) {
3148 cfg
.fix_firmwarebug
= FIX_SAMSUNG
;
3149 } else if (!strcmp(arg
, "samsung2")) {
3150 cfg
.fix_firmwarebug
= FIX_SAMSUNG2
;
3151 } else if (!strcmp(arg
, "samsung3")) {
3152 cfg
.fix_firmwarebug
= FIX_SAMSUNG3
;
3158 // check SMART status
3159 cfg
.smartcheck
= true;
3162 // check for failure of usage attributes
3163 cfg
.usagefailed
= true;
3166 // track changes in all vendor attributes
3171 // track changes in prefail vendor attributes
3175 // track changes in usage vendor attributes
3179 // track changes in SMART logs
3180 if ((arg
= strtok(NULL
, delim
)) == NULL
) {
3182 } else if (!strcmp(arg
, "selftest")) {
3183 // track changes in self-test log
3184 cfg
.selftest
= true;
3185 } else if (!strcmp(arg
, "error")) {
3186 // track changes in ATA error log
3187 cfg
.errorlog
= true;
3193 // monitor everything
3194 cfg
.smartcheck
= true;
3196 cfg
.usagefailed
= true;
3198 cfg
.selftest
= true;
3199 cfg
.errorlog
= true;
3202 // automatic offline testing enable/disable
3203 if ((arg
= strtok(NULL
, delim
)) == NULL
) {
3205 } else if (!strcmp(arg
, "on")) {
3206 cfg
.autoofflinetest
= 2;
3207 } else if (!strcmp(arg
, "off")) {
3208 cfg
.autoofflinetest
= 1;
3214 // skip disk check if in idle or standby mode
3215 if (!(arg
= strtok(NULL
, delim
)))
3218 char *endptr
= NULL
;
3219 char *next
= strchr(const_cast<char*>(arg
), ',');
3221 cfg
.powerquiet
= false;
3222 cfg
.powerskipmax
= 0;
3224 if (next
!=NULL
) *next
='\0';
3225 if (!strcmp(arg
, "never"))
3227 else if (!strcmp(arg
, "sleep"))
3229 else if (!strcmp(arg
, "standby"))
3231 else if (!strcmp(arg
, "idle"))
3236 // if optional arguments are present
3237 if (!badarg
&& next
!=NULL
) {
3239 cfg
.powerskipmax
= strtol(next
, &endptr
, 10);
3241 cfg
.powerskipmax
= 0;
3243 next
= endptr
+ (*endptr
!= '\0');
3244 if (cfg
.powerskipmax
<= 0)
3247 if (*next
!= '\0') {
3248 if (!strcmp("q", next
))
3249 cfg
.powerquiet
= true;
3258 // automatic attribute autosave enable/disable
3259 if ((arg
= strtok(NULL
, delim
)) == NULL
) {
3261 } else if (!strcmp(arg
, "on")) {
3263 } else if (!strcmp(arg
, "off")) {
3270 // warn user, and delete any previously given -s REGEXP Directives
3271 if (!cfg
.test_regex
.empty()){
3272 PrintOut(LOG_INFO
, "File %s line %d (drive %s): ignoring previous Test Directive -s %s\n",
3273 configfile
, lineno
, name
, cfg
.test_regex
.get_pattern());
3274 cfg
.test_regex
= regular_expression();
3276 // check for missing argument
3277 if (!(arg
= strtok(NULL
, delim
))) {
3282 if (!cfg
.test_regex
.compile(arg
, REG_EXTENDED
)) {
3283 // not a valid regular expression!
3284 PrintOut(LOG_CRIT
, "File %s line %d (drive %s): -s argument \"%s\" is INVALID extended regular expression. %s.\n",
3285 configfile
, lineno
, name
, arg
, cfg
.test_regex
.get_errmsg());
3289 // Do a bit of sanity checking and warn user if we think that
3290 // their regexp is "strange". User probably confused about shell
3291 // glob(3) syntax versus regular expression syntax regexp(7).
3292 if (arg
[(val
= strspn(arg
, "0123456789/.-+*|()?^$[]SLCOcnr"))])
3293 PrintOut(LOG_INFO
, "File %s line %d (drive %s): warning, character %d (%c) looks odd in extended regular expression %s\n",
3294 configfile
, lineno
, name
, val
+1, arg
[val
], arg
);
3297 // send email to address that follows
3298 if (!(arg
= strtok(NULL
,delim
)))
3301 if (!cfg
.emailaddress
.empty())
3302 PrintOut(LOG_INFO
, "File %s line %d (drive %s): ignoring previous Address Directive -m %s\n",
3303 configfile
, lineno
, name
, cfg
.emailaddress
.c_str());
3304 cfg
.emailaddress
= arg
;
3308 // email warning options
3309 if (!(arg
= strtok(NULL
, delim
)))
3311 else if (!strcmp(arg
, "once"))
3313 else if (!strcmp(arg
, "daily"))
3315 else if (!strcmp(arg
, "diminishing"))
3317 else if (!strcmp(arg
, "test"))
3319 else if (!strcmp(arg
, "exec")) {
3320 // Get the next argument (the command line)
3321 if (!(arg
= strtok(NULL
, delim
))) {
3322 PrintOut(LOG_CRIT
, "File %s line %d (drive %s): Directive %s 'exec' argument must be followed by executable path.\n",
3323 configfile
, lineno
, name
, token
);
3326 // Free the last cmd line given if any, and copy new one
3327 if (!cfg
.emailcmdline
.empty())
3328 PrintOut(LOG_INFO
, "File %s line %d (drive %s): ignoring previous mail Directive -M exec %s\n",
3329 configfile
, lineno
, name
, cfg
.emailcmdline
.c_str());
3330 cfg
.emailcmdline
= arg
;
3336 // ignore failure of usage attribute
3337 if ((val
=GetInteger(arg
=strtok(NULL
,delim
), name
, token
, lineno
, configfile
, 1, 255))<0)
3339 cfg
.monitor_attr_flags
.set(val
, MONITOR_IGN_FAILUSE
);
3342 // ignore attribute for tracking purposes
3343 if ((val
=GetInteger(arg
=strtok(NULL
,delim
), name
, token
, lineno
, configfile
, 1, 255))<0)
3345 cfg
.monitor_attr_flags
.set(val
, MONITOR_IGNORE
);
3348 // print raw value when tracking
3349 if ((val
= GetInteger(arg
=strtok(NULL
,delim
), name
, token
, lineno
, configfile
, 1, 255, excl
)) < 0)
3351 cfg
.monitor_attr_flags
.set(val
, MONITOR_RAW_PRINT
);
3352 if (*excl
== '!') // attribute change is critical
3353 cfg
.monitor_attr_flags
.set(val
, MONITOR_AS_CRIT
);
3356 // track changes in raw value (forces printing of raw value)
3357 if ((val
= GetInteger(arg
=strtok(NULL
,delim
), name
, token
, lineno
, configfile
, 1, 255, excl
)) < 0)
3359 cfg
.monitor_attr_flags
.set(val
, MONITOR_RAW_PRINT
|MONITOR_RAW
);
3360 if (*excl
== '!') // raw value change is critical
3361 cfg
.monitor_attr_flags
.set(val
, MONITOR_RAW_AS_CRIT
);
3364 // track Temperature
3365 if ((val
=Get3Integers(arg
=strtok(NULL
,delim
), name
, token
, lineno
, configfile
,
3366 &cfg
.tempdiff
, &cfg
.tempinfo
, &cfg
.tempcrit
))<0)
3370 // non-default vendor-specific attribute meaning
3371 if (!(arg
=strtok(NULL
,delim
))) {
3373 } else if (!parse_attribute_def(arg
, cfg
.attribute_defs
, PRIOR_USER
)) {
3378 // Define use of drive-specific presets.
3379 if (!(arg
= strtok(NULL
, delim
))) {
3381 } else if (!strcmp(arg
, "use")) {
3382 cfg
.ignorepresets
= false;
3383 } else if (!strcmp(arg
, "ignore")) {
3384 cfg
.ignorepresets
= true;
3385 } else if (!strcmp(arg
, "show")) {
3386 cfg
.showpresets
= true;
3387 } else if (!strcmp(arg
, "showall")) {
3394 // Directive not recognized
3395 PrintOut(LOG_CRIT
,"File %s line %d (drive %s): unknown Directive: %s\n",
3396 configfile
, lineno
, name
, token
);
3401 PrintOut(LOG_CRIT
, "File %s line %d (drive %s): Missing argument to %s Directive\n",
3402 configfile
, lineno
, name
, token
);
3405 PrintOut(LOG_CRIT
, "File %s line %d (drive %s): Invalid argument to %s Directive: %s\n",
3406 configfile
, lineno
, name
, token
, arg
);
3408 if (missingarg
|| badarg
) {
3409 PrintOut(LOG_CRIT
, "Valid arguments to %s Directive are: ", token
);
3410 printoutvaliddirectiveargs(LOG_CRIT
, sym
);
3411 PrintOut(LOG_CRIT
, "\n");
3418 // Scan directive for configuration file
3419 #define SCANDIRECTIVE "DEVICESCAN"
3421 // This is the routine that adds things to the conf_entries list.
3423 // Return values are:
3424 // 1: parsed a normal line
3425 // 0: found comment or blank line
3426 // -1: found SCANDIRECTIVE line
3427 // -2: found an error
3429 // Note: this routine modifies *line from the caller!
3430 static int ParseConfigLine(dev_config_vector
& conf_entries
, int entry
, int lineno
, /*const*/ char * line
)
3434 const char *delim
= " \n\t";
3437 // get first token: device name. If a comment, skip line
3438 if (!(name
=strtok(line
,delim
)) || *name
=='#') {
3442 // Have we detected the SCANDIRECTIVE directive?
3443 if (!strcmp(SCANDIRECTIVE
,name
)){
3446 PrintOut(LOG_INFO
,"Scan Directive %s (line %d) must be the first entry in %s\n",name
, lineno
, configfile
);
3451 // We've got a legit entry, make space to store it
3452 conf_entries
.push_back( dev_config() );
3453 dev_config
& cfg
= conf_entries
.back();
3457 // Store line number, and by default check for both device types.
3460 // parse tokens one at a time from the file.
3461 while ((token
=strtok(NULL
,delim
))){
3462 int retval
=ParseToken(token
,cfg
);
3471 PrintOut(LOG_INFO
,"Parsed token %s\n",token
);
3477 // error found on the line
3482 // If NO monitoring directives are set, then set all of them.
3483 if (!(cfg
.smartcheck
|| cfg
.usagefailed
|| cfg
.prefail
||
3484 cfg
.usage
|| cfg
.selftest
|| cfg
.errorlog
||
3485 cfg
.tempdiff
|| cfg
.tempinfo
|| cfg
.tempcrit
)) {
3487 PrintOut(LOG_INFO
,"Drive: %s, implied '-a' Directive on line %d of file %s\n",
3488 cfg
.name
.c_str(), cfg
.lineno
, configfile
);
3490 cfg
.smartcheck
= true;
3491 cfg
.usagefailed
= true;
3494 cfg
.selftest
= true;
3495 cfg
.errorlog
= true;
3498 // additional sanity check. Has user set -M options without -m?
3499 if (cfg
.emailaddress
.empty() && (!cfg
.emailcmdline
.empty() || cfg
.emailfreq
|| cfg
.emailtest
)){
3500 PrintOut(LOG_CRIT
,"Drive: %s, -M Directive(s) on line %d of file %s need -m ADDRESS Directive\n",
3501 cfg
.name
.c_str(), cfg
.lineno
, configfile
);
3505 // has the user has set <nomailer>?
3506 if (cfg
.emailaddress
== "<nomailer>") {
3507 // check that -M exec is also set
3508 if (cfg
.emailcmdline
.empty()){
3509 PrintOut(LOG_CRIT
,"Drive: %s, -m <nomailer> Directive on line %d of file %s needs -M exec Directive\n",
3510 cfg
.name
.c_str(), cfg
.lineno
, configfile
);
3513 // From here on the sign of <nomailer> is address.empty() and !cfg.emailcmdline.empty()
3514 cfg
.emailaddress
.clear();
3517 // set cfg.emailfreq to 1 (once) if user hasn't set it
3518 if ((!cfg
.emailaddress
.empty() || !cfg
.emailcmdline
.empty()) && !cfg
.emailfreq
)
3527 // Parses a configuration file. Return values are:
3528 // N=>0: found N entries
3529 // -1: syntax error in config file
3530 // -2: config file does not exist
3531 // -3: config file exists but cannot be read
3533 // In the case where the return value is 0, there are three
3535 // Empty configuration file ==> conf_entries.empty()
3536 // No configuration file ==> conf_entries[0].lineno == 0
3537 // SCANDIRECTIVE found ==> conf_entries[0].lineno != 0
3538 static int ParseConfigFile(dev_config_vector
& conf_entries
)
3540 // maximum line length in configuration file
3541 const int MAXLINELEN
= 256;
3542 // maximum length of a continued line in configuration file
3543 const int MAXCONTLINE
= 1023;
3546 // Open config file, if it exists and is not <stdin>
3547 if (!(configfile
== configfile_stdin
)) { // pointer comparison ok here
3548 if (!f
.open(configfile
,"r") && (errno
!=ENOENT
|| !configfile_alt
.empty())) {
3549 // file exists but we can't read it or it should exist due to '-c' option
3550 int ret
= (errno
!=ENOENT
? -3 : -2);
3551 PrintOut(LOG_CRIT
,"%s: Unable to open configuration file %s\n",
3552 strerror(errno
),configfile
);
3556 else // read from stdin ('-c -' option)
3559 // No configuration file found -- use fake one
3562 char fakeconfig
[] = SCANDIRECTIVE
" -a"; // TODO: Remove this hack, build cfg_entry.
3564 if (ParseConfigLine(conf_entries
, entry
, 0, fakeconfig
) != -1)
3565 throw std::logic_error("Internal error parsing "SCANDIRECTIVE
);
3570 setmode(fileno(f
), O_TEXT
); // Allow files with \r\n
3573 // configuration file exists
3574 PrintOut(LOG_INFO
,"Opened configuration file %s\n",configfile
);
3576 // parse config file line by line
3577 int lineno
= 1, cont
= 0, contlineno
= 0;
3578 char line
[MAXLINELEN
+2];
3579 char fullline
[MAXCONTLINE
+1];
3582 int len
=0,scandevice
;
3587 // make debugging simpler
3588 memset(line
,0,sizeof(line
));
3591 code
=fgets(line
, MAXLINELEN
+2, f
);
3593 // are we at the end of the file?
3596 scandevice
= ParseConfigLine(conf_entries
, entry
, contlineno
, fullline
);
3597 // See if we found a SCANDIRECTIVE directive
3600 // did we find a syntax error
3603 // the final line is part of a continuation line
3610 // input file line number
3613 // See if line is too long
3615 if (len
>MAXLINELEN
){
3617 if (line
[len
-1]=='\n')
3618 warn
="(including newline!) ";
3621 PrintOut(LOG_CRIT
,"Error: line %d of file %s %sis more than MAXLINELEN=%d characters.\n",
3622 (int)contlineno
,configfile
,warn
,(int)MAXLINELEN
);
3626 // Ignore anything after comment symbol
3627 if ((comment
=strchr(line
,'#'))){
3632 // is the total line (made of all continuation lines) too long?
3633 if (cont
+len
>MAXCONTLINE
){
3634 PrintOut(LOG_CRIT
,"Error: continued line %d (actual line %d) of file %s is more than MAXCONTLINE=%d characters.\n",
3635 lineno
, (int)contlineno
, configfile
, (int)MAXCONTLINE
);
3639 // copy string so far into fullline, and increment length
3640 strcpy(fullline
+cont
,line
);
3643 // is this a continuation line. If so, replace \ by space and look at next line
3644 if ( (lastslash
=strrchr(line
,'\\')) && !strtok(lastslash
+1," \n\t")){
3645 *(fullline
+(cont
-len
)+(lastslash
-line
))=' ';
3649 // Not a continuation line. Parse it
3650 scandevice
= ParseConfigLine(conf_entries
, entry
, contlineno
, fullline
);
3652 // did we find a scandevice directive?
3655 // did we find a syntax error
3664 // note -- may be zero if syntax of file OK, but no valid entries!
3668 /* Prints the message "=======> VALID ARGUMENTS ARE: <LIST> <=======\n", where
3669 <LIST> is the list of valid arguments for option opt. */
3670 void PrintValidArgs(char opt
) {
3673 PrintOut(LOG_CRIT
, "=======> VALID ARGUMENTS ARE: ");
3674 if (!(s
= GetValidArgList(opt
)))
3675 PrintOut(LOG_CRIT
, "Error constructing argument list for option %c", opt
);
3677 PrintOut(LOG_CRIT
, "%s", (char *)s
);
3678 PrintOut(LOG_CRIT
, " <=======\n");
3681 // Return true if absolute path name
3682 static bool is_abs_path(const char * path
)
3686 #if defined(_WIN32) || defined(__CYGWIN__)
3690 sscanf(path
, "%*1[A-Za-z]:%*1[/\\]%n", &n
);
3697 // Parses input line, prints usage message and
3698 // version/license/copyright messages
3699 void ParseOpts(int argc
, char **argv
){
3703 // Please update GetValidArgList() if you edit shortopts
3704 const char *shortopts
= "c:l:q:dDni:p:r:s:A:B:Vh?";
3706 // Please update GetValidArgList() if you edit longopts
3707 struct option longopts
[] = {
3708 { "configfile", required_argument
, 0, 'c' },
3709 { "logfacility", required_argument
, 0, 'l' },
3710 { "quit", required_argument
, 0, 'q' },
3711 { "debug", no_argument
, 0, 'd' },
3712 { "showdirectives", no_argument
, 0, 'D' },
3713 { "interval", required_argument
, 0, 'i' },
3715 { "no-fork", no_argument
, 0, 'n' },
3717 { "pidfile", required_argument
, 0, 'p' },
3718 { "report", required_argument
, 0, 'r' },
3719 { "savestates", required_argument
, 0, 's' },
3720 { "attributelog", required_argument
, 0, 'A' },
3721 { "drivedb", required_argument
, 0, 'B' },
3722 #if defined(_WIN32) || defined(__CYGWIN__)
3723 { "service", no_argument
, 0, 'n' },
3725 { "version", no_argument
, 0, 'V' },
3726 { "license", no_argument
, 0, 'V' },
3727 { "copyright", no_argument
, 0, 'V' },
3728 { "help", no_argument
, 0, 'h' },
3729 { "usage", no_argument
, 0, 'h' },
3734 bool badarg
= false;
3735 bool no_defaultdb
= false; // set true on '-B FILE'
3737 // Parse input options. This horrible construction is so that emacs
3738 // indents properly. Sorry.
3739 while (-1 != (optchar
=
3740 getopt_long(argc
, argv
, shortopts
, longopts
, NULL
)
3746 if (!(strcmp(optarg
,"nodev"))) {
3748 } else if (!(strcmp(optarg
,"nodevstartup"))) {
3750 } else if (!(strcmp(optarg
,"never"))) {
3752 } else if (!(strcmp(optarg
,"onecheck"))) {
3755 } else if (!(strcmp(optarg
,"showtests"))) {
3758 } else if (!(strcmp(optarg
,"errors"))) {
3765 // set the log facility level
3766 if (!strcmp(optarg
, "daemon"))
3767 facility
=LOG_DAEMON
;
3768 else if (!strcmp(optarg
, "local0"))
3769 facility
=LOG_LOCAL0
;
3770 else if (!strcmp(optarg
, "local1"))
3771 facility
=LOG_LOCAL1
;
3772 else if (!strcmp(optarg
, "local2"))
3773 facility
=LOG_LOCAL2
;
3774 else if (!strcmp(optarg
, "local3"))
3775 facility
=LOG_LOCAL3
;
3776 else if (!strcmp(optarg
, "local4"))
3777 facility
=LOG_LOCAL4
;
3778 else if (!strcmp(optarg
, "local5"))
3779 facility
=LOG_LOCAL5
;
3780 else if (!strcmp(optarg
, "local6"))
3781 facility
=LOG_LOCAL6
;
3782 else if (!strcmp(optarg
, "local7"))
3783 facility
=LOG_LOCAL7
;
3788 // enable debug mode
3793 #ifndef _WIN32 // On Windows, --service is already handled by daemon_main()
3798 // print summary of all valid directives
3804 // Period (time interval) for checking
3805 // strtol will set errno in the event of overflow, so we'll check it.
3807 lchecktime
= strtol(optarg
, &tailptr
, 10);
3808 if (*tailptr
!= '\0' || lchecktime
< 10 || lchecktime
> INT_MAX
|| errno
) {
3811 PrintOut(LOG_CRIT
, "======> INVALID INTERVAL: %s <=======\n", optarg
);
3812 PrintOut(LOG_CRIT
, "======> INTERVAL MUST BE INTEGER BETWEEN %d AND %d <=======\n", 10, INT_MAX
);
3813 PrintOut(LOG_CRIT
, "\nUse smartd -h to get a usage summary\n\n");
3816 checktime
= (int)lchecktime
;
3819 // report IOCTL transactions
3824 // split_report_arg() may modify its first argument string, so use a
3825 // copy of optarg in case we want optarg for an error message.
3826 if (!(s
= strdup(optarg
))) {
3827 PrintOut(LOG_CRIT
, "No memory to process -r option - exiting\n");
3830 if (split_report_arg(s
, &i
)) {
3832 } else if (i
<1 || i
>3) {
3835 PrintOut(LOG_CRIT
, "======> INVALID REPORT LEVEL: %s <=======\n", optarg
);
3836 PrintOut(LOG_CRIT
, "======> LEVEL MUST BE INTEGER BETWEEN 1 AND 3<=======\n");
3838 } else if (!strcmp(s
,"ioctl")) {
3839 con
->reportataioctl
= con
->reportscsiioctl
= i
;
3840 } else if (!strcmp(s
,"ataioctl")) {
3841 con
->reportataioctl
= i
;
3842 } else if (!strcmp(s
,"scsiioctl")) {
3843 con
->reportscsiioctl
= i
;
3847 free(s
); // TODO: use std::string
3851 // alternate configuration file
3852 if (strcmp(optarg
,"-"))
3853 configfile
= (configfile_alt
= optarg
).c_str();
3854 else // read from stdin
3855 configfile
=configfile_stdin
;
3858 // output file with PID number
3862 // path prefix of persistent state file
3863 state_path_prefix
= optarg
;
3866 // path prefix of attribute log file
3867 attrlog_path_prefix
= optarg
;
3871 const char * path
= optarg
;
3872 if (*path
== '+' && path
[1])
3875 no_defaultdb
= true;
3876 unsigned char savedebug
= debugmode
; debugmode
= 1;
3877 if (!read_drive_database(path
))
3879 debugmode
= savedebug
;
3883 // print version and CVS info
3885 PrintOut(LOG_INFO
, "%s", format_version_info("smartd", true /*full*/).c_str());
3889 // help: print summary of command-line options
3897 // unrecognized option
3900 // Point arg to the argument in which this option was found.
3901 arg
= argv
[optind
-1];
3902 // Check whether the option is a long option that doesn't map to -h.
3903 if (arg
[1] == '-' && optchar
!= 'h') {
3904 // Iff optopt holds a valid option then argument must be missing.
3905 if (optopt
&& (strchr(shortopts
, optopt
) != NULL
)) {
3906 PrintOut(LOG_CRIT
, "=======> ARGUMENT REQUIRED FOR OPTION: %s <=======\n",arg
+2);
3907 PrintValidArgs(optopt
);
3909 PrintOut(LOG_CRIT
, "=======> UNRECOGNIZED OPTION: %s <=======\n\n",arg
+2);
3911 PrintOut(LOG_CRIT
, "\nUse smartd --help to get a usage summary\n\n");
3915 // Iff optopt holds a valid option then argument must be missing.
3916 if (strchr(shortopts
, optopt
) != NULL
){
3917 PrintOut(LOG_CRIT
, "=======> ARGUMENT REQUIRED FOR OPTION: %c <=======\n",optopt
);
3918 PrintValidArgs(optopt
);
3920 PrintOut(LOG_CRIT
, "=======> UNRECOGNIZED OPTION: %c <=======\n\n",optopt
);
3922 PrintOut(LOG_CRIT
, "\nUse smartd -h to get a usage summary\n\n");
3929 // Check to see if option had an unrecognized or incorrect argument.
3933 // It would be nice to print the actual option name given by the user
3934 // here, but we just print the short form. Please fix this if you know
3935 // a clean way to do it.
3936 PrintOut(LOG_CRIT
, "=======> INVALID ARGUMENT TO -%c: %s <======= \n", optchar
, optarg
);
3937 PrintValidArgs(optchar
);
3938 PrintOut(LOG_CRIT
, "\nUse smartd -h to get a usage summary\n\n");
3943 // non-option arguments are not allowed
3944 if (argc
> optind
) {
3947 PrintOut(LOG_CRIT
, "=======> UNRECOGNIZED ARGUMENT: %s <=======\n\n", argv
[optind
]);
3948 PrintOut(LOG_CRIT
, "\nUse smartd -h to get a usage summary\n\n");
3952 // no pidfile in debug mode
3953 if (debugmode
&& !pid_file
.empty()) {
3956 PrintOut(LOG_CRIT
, "=======> INVALID CHOICE OF OPTIONS: -d and -p <======= \n\n");
3957 PrintOut(LOG_CRIT
, "Error: pid file %s not written in debug (-d) mode\n\n", pid_file
.c_str());
3961 // absolute path is required due to chdir('/') after fork().
3962 if (!state_path_prefix
.empty() && !debugmode
&& !is_abs_path(state_path_prefix
.c_str())) {
3965 PrintOut(LOG_CRIT
, "=======> INVALID CHOICE OF OPTIONS: -s <======= \n\n");
3966 PrintOut(LOG_CRIT
, "Error: relative path %s is only allowed in debug (-d) mode\n\n",
3967 state_path_prefix
.c_str());
3971 // absolute path is required due to chdir('/') after fork().
3972 if (!attrlog_path_prefix
.empty() && !debugmode
&& !is_abs_path(attrlog_path_prefix
.c_str())) {
3975 PrintOut(LOG_CRIT
, "=======> INVALID CHOICE OF OPTIONS: -s <======= \n\n");
3976 PrintOut(LOG_CRIT
, "Error: relative path %s is only allowed in debug (-d) mode\n\n",
3977 attrlog_path_prefix
.c_str());
3981 // Read or init drive database
3982 if (!no_defaultdb
) {
3983 unsigned char savedebug
= debugmode
; debugmode
= 1;
3984 if (!read_default_drive_databases())
3986 debugmode
= savedebug
;
3993 // Function we call if no configuration file was found or if the
3994 // SCANDIRECTIVE Directive was found. It makes entries for device
3995 // names returned by scan_smart_devices() in os_OSNAME.cpp
3996 static int MakeConfigEntries(const dev_config
& base_cfg
,
3997 dev_config_vector
& conf_entries
, smart_device_list
& scanned_devs
, const char * type
)
3999 // make list of devices
4000 smart_device_list devlist
;
4001 if (!smi()->scan_smart_devices(devlist
, (*type
? type
: 0)))
4002 PrintOut(LOG_CRIT
,"Problem creating device name scan list\n");
4004 // if no devices, or error constructing list, return
4005 if (devlist
.size() <= 0)
4008 // loop over entries to create
4009 for (unsigned i
= 0; i
< devlist
.size(); i
++) {
4010 // Move device pointer
4011 smart_device
* dev
= devlist
.release(i
);
4012 scanned_devs
.push_back(dev
);
4014 // Copy configuration, update device and type name
4015 conf_entries
.push_back(base_cfg
);
4016 dev_config
& cfg
= conf_entries
.back();
4017 cfg
.name
= dev
->get_info().info_name
;
4018 cfg
.dev_type
= type
;
4021 return devlist
.size();
4024 static void CanNotRegister(const char *name
, const char *type
, int line
, bool scandirective
)
4026 if (!debugmode
&& scandirective
)
4029 PrintOut(scandirective
?LOG_INFO
:LOG_CRIT
,
4030 "Unable to register %s device %s at line %d of file %s\n",
4031 type
, name
, line
, configfile
);
4033 PrintOut(LOG_INFO
,"Unable to register %s device %s\n",
4038 // Returns negative value (see ParseConfigFile()) if config file
4039 // had errors, else number of entries which may be zero or positive.
4040 // If we found no configuration file, or it contained SCANDIRECTIVE,
4041 // then *scanning is set to 1, else 0.
4042 static int ReadOrMakeConfigEntries(dev_config_vector
& conf_entries
, smart_device_list
& scanned_devs
)
4044 // parse configuration file configfile (normally /etc/smartd.conf)
4045 int entries
= ParseConfigFile(conf_entries
);
4048 // There was an error reading the configuration file.
4049 conf_entries
.clear();
4051 PrintOut(LOG_CRIT
, "Configuration file %s has fatal syntax errors.\n", configfile
);
4055 // no error parsing config file.
4057 // we did not find a SCANDIRECTIVE and did find valid entries
4058 PrintOut(LOG_INFO
, "Configuration file %s parsed.\n", configfile
);
4060 else if (conf_entries
.size() == 1) {
4061 // we found a SCANDIRECTIVE or there was no configuration file so
4062 // scan. Configuration file's first entry contains all options
4064 dev_config first
= conf_entries
.front();
4065 conf_entries
.clear();
4068 PrintOut(LOG_INFO
,"Configuration file %s was parsed, found %s, scanning devices\n", configfile
, SCANDIRECTIVE
);
4070 PrintOut(LOG_INFO
,"No configuration file %s found, scanning devices\n", configfile
);
4072 // make config list of devices to search for
4073 MakeConfigEntries(first
, conf_entries
, scanned_devs
, first
.dev_type
.c_str());
4075 // warn user if scan table found no devices
4076 if (conf_entries
.empty())
4077 PrintOut(LOG_CRIT
,"In the system's table of devices NO devices found to scan\n");
4080 PrintOut(LOG_CRIT
,"Configuration file %s parsed but has no entries (like /dev/hda)\n",configfile
);
4082 return conf_entries
.size();
4086 // This function tries devices from conf_entries. Each one that can be
4087 // registered is moved onto the [ata|scsi]devices lists and removed
4088 // from the conf_entries list.
4089 static void RegisterDevices(const dev_config_vector
& conf_entries
, smart_device_list
& scanned_devs
,
4090 dev_config_vector
& configs
, dev_state_vector
& states
, smart_device_list
& devices
)
4092 // start by clearing lists/memory of ALL existing devices
4098 for (unsigned i
= 0; i
< conf_entries
.size(); i
++){
4100 dev_config cfg
= conf_entries
[i
];
4102 // get device of appropriate type
4103 smart_device_auto_ptr dev
;
4104 bool scanning
= false;
4106 // Device may already be detected during devicescan
4107 if (i
< scanned_devs
.size()) {
4108 dev
= scanned_devs
.release(i
);
4114 dev
= smi()->get_smart_device(cfg
.name
.c_str(), cfg
.dev_type
.c_str());
4116 if (cfg
.dev_type
.empty())
4117 PrintOut(LOG_INFO
,"Device: %s, unable to autodetect device type\n", cfg
.name
.c_str());
4119 PrintOut(LOG_INFO
,"Device: %s, unsupported device type '%s'\n", cfg
.name
.c_str(), cfg
.dev_type
.c_str());
4125 smart_device::device_info oldinfo
= dev
->get_info();
4127 // Open with autodetect support, may return 'better' device
4128 dev
.replace( dev
->autodetect_open() );
4130 // Report if type has changed
4131 if (oldinfo
.dev_type
!= dev
->get_dev_type())
4132 PrintOut(LOG_INFO
,"Device: %s, type changed from '%s' to '%s'\n",
4133 cfg
.name
.c_str(), oldinfo
.dev_type
.c_str(), dev
->get_dev_type());
4135 if (!dev
->is_open()) {
4136 // For linux+devfs, a nonexistent device gives a strange error
4137 // message. This makes the error message a bit more sensible.
4138 // If no debug and scanning - don't print errors
4139 if (debugmode
|| !scanning
)
4140 PrintOut(LOG_INFO
, "Device: %s, open() failed: %s\n", dev
->get_info_name(), dev
->get_errmsg());
4144 // Update informal name
4145 cfg
.name
= dev
->get_info().info_name
;
4146 PrintOut(LOG_INFO
, "Device: %s, opened\n", cfg
.name
.c_str());
4148 // Prepare initial state
4151 // register ATA devices
4153 if (ATADeviceScan(cfg
, state
, dev
->to_ata())) {
4154 CanNotRegister(cfg
.name
.c_str(), "ATA", cfg
.lineno
, scanning
);
4158 // or register SCSI devices
4159 else if (dev
->is_scsi()){
4160 if (SCSIDeviceScan(cfg
, state
, dev
->to_scsi())) {
4161 CanNotRegister(cfg
.name
.c_str(), "SCSI", cfg
.lineno
, scanning
);
4166 PrintOut(LOG_INFO
, "Device: %s, neither ATA nor SCSI device\n", cfg
.name
.c_str());
4171 // move onto the list of devices
4172 configs
.push_back(cfg
);
4173 states
.push_back(state
);
4174 devices
.push_back(dev
);
4176 // if device is explictly listed and we can't register it, then
4177 // exit unless the user has specified that the device is removable
4178 else if (!scanning
) {
4179 if (cfg
.removable
|| quit
==2)
4180 PrintOut(LOG_INFO
, "Device %s not available\n", cfg
.name
.c_str());
4182 PrintOut(LOG_CRIT
, "Unable to register device %s (no Directive -d removable). Exiting.\n", cfg
.name
.c_str());
4190 // Main program without exception handling
4191 int main_worker(int argc
, char **argv
)
4193 // Initialize interface
4194 smart_interface::init();
4198 // external control variables for ATA disks
4199 smartmonctrl control
;
4201 // is it our first pass through?
4202 bool firstpass
= true;
4204 // next time to wake up
4207 // for simplicity, null all global communications variables/lists
4209 memset(con
, 0,sizeof(control
));
4211 // parse input and print header and usage info if needed
4212 ParseOpts(argc
,argv
);
4214 // do we mute printing from ataprint commands?
4215 con
->printing_switchable
= false;
4216 con
->dont_print
= !debugmode
;
4218 // Configuration for each device
4219 dev_config_vector configs
;
4221 dev_state_vector states
;
4222 // Devices to monitor
4223 smart_device_list devices
;
4225 bool write_states_always
= true;
4227 // the main loop of the code
4230 // are we exiting from a signal?
4231 if (caughtsigEXIT
) {
4232 // are we exiting with SIGTERM?
4233 int isterm
=(caughtsigEXIT
==SIGTERM
);
4234 int isquit
=(caughtsigEXIT
==SIGQUIT
);
4235 int isok
=debugmode
?isterm
|| isquit
:isterm
;
4237 PrintOut(isok
?LOG_INFO
:LOG_CRIT
, "smartd received signal %d: %s\n",
4238 caughtsigEXIT
, strsignal(caughtsigEXIT
));
4243 // Write state files
4244 if (!state_path_prefix
.empty())
4245 write_all_dev_states(configs
, states
);
4250 // Should we (re)read the config file?
4251 if (firstpass
|| caughtsigHUP
){
4254 // Workaround for missing SIGQUIT via keyboard on Cygwin
4255 if (caughtsigHUP
==2) {
4256 // Simulate SIGQUIT if another SIGINT arrives soon
4259 if (caughtsigHUP
==2) {
4260 caughtsigEXIT
=SIGQUIT
;
4266 // Write state files
4267 if (!state_path_prefix
.empty())
4268 write_all_dev_states(configs
, states
);
4272 "Signal HUP - rereading configuration file %s\n":
4273 "\a\nSignal INT - rereading configuration file %s ("SIGQUIT_KEYNAME
" quits)\n\n",
4278 dev_config_vector conf_entries
; // Entries read from smartd.conf
4279 smart_device_list scanned_devs
; // Devices found during scan
4280 // (re)reads config file, makes >=0 entries
4281 int entries
= ReadOrMakeConfigEntries(conf_entries
, scanned_devs
);
4284 // checks devices, then moves onto ata/scsi list or deallocates.
4285 RegisterDevices(conf_entries
, scanned_devs
, configs
, states
, devices
);
4286 if (!(configs
.size() == devices
.size() && configs
.size() == states
.size()))
4287 throw std::logic_error("Invalid result from RegisterDevices");
4289 else if (quit
==2 || ((quit
==0 || quit
==1) && !firstpass
)) {
4290 // user has asked to continue on error in configuration file
4292 PrintOut(LOG_INFO
,"Reusing previous configuration\n");
4295 // exit with configuration file error status
4296 int status
= (entries
==-3 ? EXIT_READCONF
: entries
==-2 ? EXIT_NOCONF
: EXIT_BADCONF
);
4301 // Log number of devices we are monitoring...
4302 if (devices
.size() > 0 || quit
==2 || (quit
==1 && !firstpass
)) {
4304 for (unsigned i
= 0; i
< devices
.size(); i
++) {
4305 if (devices
.at(i
)->is_ata())
4308 PrintOut(LOG_INFO
,"Monitoring %d ATA and %d SCSI devices\n",
4309 numata
, devices
.size() - numata
);
4312 PrintOut(LOG_INFO
,"Unable to monitor any SMART enabled devices. Try debug (-d) option. Exiting...\n");
4317 // user has asked to print test schedule
4318 PrintTestSchedule(configs
, states
, devices
);
4325 // Always write state files after (re)configuration
4326 write_states_always
= true;
4329 // check all devices once,
4330 // self tests are not started in first pass unless '-q onecheck' is specified
4331 CheckDevicesOnce(configs
, states
, devices
, (!firstpass
|| quit
==3));
4333 // Write state files
4334 if (!state_path_prefix
.empty())
4335 write_all_dev_states(configs
, states
, write_states_always
);
4336 write_states_always
= false;
4338 // Write attribute logs
4339 if (!attrlog_path_prefix
.empty())
4340 write_all_dev_attrlogs(configs
, states
);
4342 // user has asked us to exit after first check
4344 PrintOut(LOG_INFO
,"Started with '-q onecheck' option. All devices sucessfully checked once.\n"
4345 "smartd is exiting (exit status 0)\n");
4349 // fork into background if needed
4350 if (firstpass
&& !debugmode
) {
4354 // set exit and signal handlers, write PID file, set wake-up time
4356 Initialize(&wakeuptime
);
4360 // sleep until next check time, or a signal arrives
4361 wakeuptime
= dosleep(wakeuptime
, write_states_always
);
4368 int main(int argc
, char **argv
)
4370 // Windows: internal main function started direct or by service control manager
4371 static int smartd_main(int argc
, char **argv
)
4376 // Do the real work ...
4377 status
= main_worker(argc
, argv
);
4380 // EXIT(status) arrives here
4383 catch (const std::bad_alloc
& /*ex*/) {
4384 // Memory allocation failed (also thrown by std::operator new)
4385 PrintOut(LOG_CRIT
, "Smartd: Out of memory\n");
4386 status
= EXIT_NOMEM
;
4388 catch (const std::exception
& ex
) {
4389 // Other fatal errors
4390 PrintOut(LOG_CRIT
, "Smartd: Exception: %s\n", ex
.what());
4391 status
= EXIT_BADCODE
;
4395 status
= Goodbye(status
);
4398 daemon_winsvc_exitcode
= status
;
4405 // Main function for Windows
4406 int main(int argc
, char **argv
){
4407 // Options for smartd windows service
4408 static const daemon_winsvc_options svc_opts
= {
4409 "--service", // cmd_opt
4410 "smartd", "SmartD Service", // servicename, displayname
4412 "Controls and monitors storage devices using the Self-Monitoring, "
4413 "Analysis and Reporting Technology System (S.M.A.R.T.) "
4414 "built into ATA and SCSI Hard Drives. "
4417 // daemon_main() handles daemon and service specific commands
4418 // and starts smartd_main() direct, from a new process,
4419 // or via service control manager
4420 return daemon_main("smartd", &svc_opts
, smartd_main
, argc
, argv
);