]> git.ipfire.org Git - thirdparty/util-linux.git/blame_incremental - sys-utils/hwclock.8.in
fixed mount man page typo, "bythe" -> "by the"
[thirdparty/util-linux.git] / sys-utils / hwclock.8.in
... / ...
CommitLineData
1.\" hwclock.8.in -- man page for util-linux' hwclock
2.\"
3.\" 2015-01-07 J William Piggott
4.\" Authored new section: DATE-TIME CONFIGURATION.
5.\" Subsections: Keeping Time..., LOCAL vs UTC, POSIX vs 'RIGHT'.
6.\"
7.TH HWCLOCK 8 "July 2017" "util-linux" "System Administration"
8.SH NAME
9hwclock \- time clocks utility
10.SH SYNOPSIS
11.B hwclock
12.RI [ function ]
13.RI [ option ...]
14.
15.SH DESCRIPTION
16.B hwclock
17is an administration tool for the time clocks. It can: display the
18Hardware Clock time; set the Hardware Clock to a specified time; set the
19Hardware Clock from the System Clock; set the System Clock from the
20Hardware Clock; compensate for Hardware Clock drift; correct the System
21Clock timescale; set the kernel's timezone, NTP timescale, and epoch
22(Alpha only); and predict future
23Hardware Clock values based on its drift rate.
24.PP
25Since v2.26 important changes were made to the
26.B \-\-hctosys
27function and the
28.B \-\-directisa
29option, and a new option
30.B \-\-update\-drift
31was added. See their respective descriptions below.
32.
33.SH FUNCTIONS
34The following functions are mutually exclusive, only one can be given at
35a time. If none is given, the default is \fB\-\-show\fR.
36.TP
37.B \-a, \-\-adjust
38Add or subtract time from the Hardware Clock to account for systematic
39drift since the last time the clock was set or adjusted. See the
40discussion below, under
41.BR "The Adjust Function" .
42.
43.TP
44.B \-\-getepoch
45.TQ
46.B \-\-setepoch
47These functions are for Alpha machines only, and are only available
48through the Linux kernel RTC driver.
49.sp
50They are used to read and set the kernel's Hardware Clock epoch value.
51Epoch is the number of years into AD to which a zero year value in the
52Hardware Clock refers. For example, if the machine's BIOS sets the year
53counter in the Hardware Clock to contain the number of full years since
541952, then the kernel's Hardware Clock epoch value must be 1952.
55.sp
56The \fB\%\-\-setepoch\fR function requires using the
57.B \%\-\-epoch
58option to specify the year. For example:
59.RS
60.IP "" 4
61.B hwclock\ \-\-setepoch\ \-\-epoch=1952
62.PP
63The RTC driver attempts to guess the correct epoch value, so setting it
64may not be required.
65.PP
66This epoch value is used whenever
67.B \%hwclock
68reads or sets the Hardware Clock on an Alpha machine. For ISA machines
69the kernel uses the fixed Hardware Clock epoch of 1900.
70.RE
71.
72.TP
73.B \-\-predict
74Predict what the Hardware Clock will read in the future based upon the
75time given by the
76.B \-\-date
77option and the information in
78.IR @ADJTIME_PATH@ .
79This is useful, for example, to account for drift when setting a
80Hardware Clock wakeup (aka alarm). See
81.BR \%rtcwake (8).
82.sp
83Do not use this function if the Hardware Clock is being modified by
84anything other than the current operating system's
85.B \%hwclock
86command, such as \%'11\ minute\ mode' or from dual-booting another OS.
87.
88.TP
89.BR \-r , \ \-\-show
90.TQ
91.B \-\-get
92.br
93Read the Hardware Clock and print its time to standard output in the
94.B ISO 8601
95format.
96The time shown is always in local time, even if you keep your Hardware Clock
97in UTC. See the
98.B \%\-\-localtime
99option.
100.sp
101Showing the Hardware Clock time is the default when no function is specified.
102.sp
103The
104.B \-\-get
105function also applies drift correction to the time read, based upon the
106information in
107.IR @ADJTIME_PATH@ .
108Do not use this function if the Hardware Clock is being modified by
109anything other than the current operating system's
110.B \%hwclock
111command, such as \%'11\ minute\ mode' or from dual-booting another OS.
112.
113.TP
114.BR \-s , \ \-\-hctosys
115Set the System Clock from the Hardware Clock. The time read from the Hardware
116Clock is compensated to account for systematic drift before using it to set the
117System Clock. See the discussion below, under
118.BR "The Adjust Function" .
119.sp
120The System Clock must be kept in the UTC timescale for date-time
121applications to work correctly in conjunction with the timezone configured
122for the system. If the Hardware Clock is kept in local time then the time read
123from it must be shifted to the UTC timescale before using it to set the System
124Clock. The
125.B \%\-\-hctosys
126function does this based upon the information in the
127.I @ADJTIME_PATH@
128file or the command line arguments
129.BR \%\-\-localtime " and " \-\-utc .
130Note: no daylight saving adjustment is made. See the discussion below, under
131.BR "LOCAL vs UTC" .
132.sp
133The kernel also keeps a timezone value, the
134.B \%\-\-hctosys
135function sets it to the timezone configured for the system. The system
136timezone is configured by the TZ environment variable or the
137.I \%/etc/localtime
138file, as
139.BR \%tzset (3)
140would interpret them.
141The obsolete tz_dsttime field of the kernel's timezone value is set
142to zero. (For details on what this field used to mean, see
143.BR \%settimeofday (2).)
144.sp
145When used in a startup script, making the
146.B \%\-\-hctosys
147function the first caller of
148.BR \%settimeofday (2)
149from boot, it will set the NTP \%'11\ minute\ mode' timescale via the
150.I \%persistent_clock_is_local
151kernel variable. If the Hardware Clock's timescale configuration is
152changed then a reboot is required to inform the kernel. See the
153discussion below, under
154.BR "Automatic Hardware Clock Synchronization by the Kernel" .
155.sp
156This is a good function to use in one of the system startup scripts before the
157file systems are mounted read/write.
158.sp
159This function should never be used on a running system. Jumping system time
160will cause problems, such as corrupted filesystem timestamps. Also, if
161something has changed the Hardware Clock, like NTP's \%'11\ minute\ mode', then
162.B \%\-\-hctosys
163will set the time incorrectly by including drift compensation.
164.sp
165Drift compensation can be inhibited by setting the drift factor in
166.I @ADJTIME_PATH@
167to zero. This setting will be persistent as long as the
168.BR \%\-\-update\-drift " option is not used with " \%\-\-systohc
169at shutdown (or anywhere else). Another way to inhibit this is by using the
170.BR \%\-\-noadjfile " option when calling the " \%\-\-hctosys
171function. A third method is to delete the
172.IR @ADJTIME_PATH@ " file."
173.B Hwclock
174will then default to using the UTC timescale for the Hardware Clock. If
175the Hardware Clock is ticking local time it will need to be defined in
176the file. This can be done by calling
177.BR hwclock\ \-\-localtime\ \-\-adjust ;
178when the file is not present this command will not actually
179adjust the Clock, but it will create the file with local time
180configured, and a drift factor of zero.
181.sp
182A condition under which inhibiting
183.BR hwclock 's
184drift correction may be desired is when dual-booting multiple operating
185systems. If while this instance of Linux is stopped, another OS changes
186the Hardware Clock's value, then when this instance is started again the
187drift correction applied will be incorrect.
188.sp
189.RB "For " hwclock 's
190drift correction to work properly it is imperative that nothing changes
191the Hardware Clock while its Linux instance is not running.
192.
193.TP
194.B \-\-set
195Set the Hardware Clock to the time given by the
196.B \-\-date
197option, and update the timestamps in
198.IR @ADJTIME_PATH@ .
199With the
200.B \%\-\-update-drift
201option also (re)calculate the drift factor. Try it without the option if
202.BR \%\-\-set " fails. See " \%\-\-update-drift " below."
203.
204.TP
205.B \-\-systz
206This is an alternate to the
207.B \%\-\-hctosys
208function that does not read the Hardware Clock nor set the System Clock;
209consequently there is not any drift correction. It is intended to be
210used in a startup script on systems with kernels above version 2.6 where
211you know the System Clock has been set from the Hardware Clock by the
212kernel during boot.
213.sp
214It does the following things that are detailed above in the
215.BR \%\-\-hctosys " function:"
216.RS
217.IP \(bu 2
218Corrects the System Clock timescale to UTC as needed. Only instead of
219accomplishing this by setting the System Clock,
220.B hwclock
221simply informs the kernel and it handles the change.
222.IP \(bu 2
223Sets the kernel's NTP \%'11\ minute\ mode' timescale.
224.IP \(bu 2
225Sets the kernel's timezone.
226.PP
227The first two are only available on the first call of
228.BR \%settimeofday (2)
229after boot. Consequently this option only makes sense when used in a
230startup script. If the Hardware Clocks timescale configuration is
231changed then a reboot would be required to inform the kernel.
232.RE
233.
234.TP
235.BR \-w , \ \-\-systohc
236Set the Hardware Clock from the System Clock, and update the timestamps in
237.IR @ADJTIME_PATH@ .
238With the
239.B \%\-\-update-drift
240option also (re)calculate the drift factor. Try it without the option if
241.BR \%\-\-systohc " fails. See " \%\-\-update-drift " below."
242.
243.TP
244.BR \-V , \ \-\-version
245Display version information and exit.
246.
247.TP
248.BR \-h , \ \-\-help
249Display help text and exit.
250.
251.SH OPTIONS
252.
253.TP
254.BI \-\-adjfile= filename
255.RI "Override the default " @ADJTIME_PATH@ " file path."
256.
257.TP
258.BI \%\-\-date= date_string
259This option must be used with the
260.B \-\-set
261or
262.B \%\-\-predict
263functions, otherwise it is ignored.
264.RS
265.IP "" 4
266.B "hwclock\ \-\-set\ \-\-date='16:45'"
267.IP "" 4
268.B "hwclock\ \-\-predict\ \-\-date='2525-08-14\ 07:11:05'"
269.PP
270The argument must be in local time, even if you keep your Hardware Clock in
271UTC. See the
272.B \%\-\-localtime
273option. Therefore, the argument should not include any timezone information.
274It also should not be a relative time like "+5 minutes", because
275.BR \%hwclock 's
276precision depends upon correlation between the argument's value and when the
277enter key is pressed. Fractional seconds are silently dropped. This option is
278capable of understanding many time and date formats, but the previous
279parameters should be observed.
280.RE
281.
282.TP
283.BI \%\-\-delay= seconds
284This option allows to overwrite internally used delay when set clock time. The
285default is 0.5 (500ms) for rtc_cmos, for another RTC types the delay is 0. If
286RTC type is impossible to determine (from sysfs) then it defaults also to 0.5
287to be backwardly compatible.
288.RS
289.PP
290The 500ms default is based on commonly used MC146818A-compatible (x86) hardware clock. This
291Hardware Clock can only be set to any integer time plus one half second. The
292integer time is required because there is no interface to set or get a
293fractional second. The additional half second delay is because the Hardware
294Clock updates to the following second precisely 500 ms after setting the new
295time. Unfortunately, this behavior is hardware specific and in same cases
296another delay is required.
297.RE
298.
299.TP
300.TP
301.BR \-D ", " \-\-debug
302.RB Use\ \-\-verbose .
303.RB The\ \%\-\-debug\ option
304has been deprecated and may be repurposed or removed in a future release.
305.
306.TP
307.B \-\-directisa
308This option is meaningful for ISA compatible machines in the x86 and
309x86_64 family. For other machines, it has no effect. This option tells
310.B \%hwclock
311to use explicit I/O instructions to access the Hardware Clock.
312Without this option,
313.B \%hwclock
314will use the rtc device file, which it assumes to be driven by the Linux
315RTC device driver. As of v2.26 it will no longer automatically use
316directisa when the rtc driver is unavailable; this was causing an unsafe
317condition that could allow two processes to access the Hardware Clock at
318the same time. Direct hardware access from userspace should only be
319used for testing, troubleshooting, and as a last resort when all other
320methods fail. See the
321.BR \-\-rtc " option."
322.
323.TP
324.BI \-\-epoch= year
325This option is required when using the
326.BR \%\-\-setepoch \ function.
327.RI "The minimum " year
328value is 1900. The maximum is system dependent
329.RB ( ULONG_MAX\ -\ 1 ).
330.
331.TP
332.BR \-f , \ \-\-rtc=\fIfilename\fR
333.RB "Override " \%hwclock 's
334default rtc device file name. Otherwise it will
335use the first one found in this order:
336.in +4
337.br
338.I /dev/rtc0
339.br
340.I /dev/rtc
341.br
342.I /dev/misc/rtc
343.br
344.in
345.RB "For " IA-64:
346.in +4
347.br
348.I /dev/efirtc
349.br
350.I /dev/misc/efirtc
351.in
352.
353.TP
354.BR \-l , \ \-\-localtime
355.TQ
356.BR \-u ", " \-\-utc
357Indicate which timescale the Hardware Clock is set to.
358.sp
359The Hardware Clock may be configured to use either the UTC or the local
360timescale, but nothing in the clock itself says which alternative is
361being used. The
362.BR \%\-\-localtime " or " \-\-utc
363options give this information to the
364.B \%hwclock
365command. If you specify the wrong one (or specify neither and take a
366wrong default), both setting and reading the Hardware Clock will be
367incorrect.
368.sp
369If you specify neither
370.BR \-\-utc " nor " \%\-\-localtime
371then the one last given with a set function
372.RB ( \-\-set ", " \%\-\-systohc ", or " \%\-\-adjust ),
373as recorded in
374.IR @ADJTIME_PATH@ ,
375will be used. If the adjtime file doesn't exist, the default is UTC.
376.sp
377Note: daylight saving time changes may be inconsistent when the
378Hardware Clock is kept in local time. See the discussion below, under
379.BR "LOCAL vs UTC" .
380.
381.TP
382.B \-\-noadjfile
383Disable the facilities provided by
384.IR @ADJTIME_PATH@ .
385.B \%hwclock
386will not read nor write to that file with this option. Either
387.BR \-\-utc " or " \%\-\-localtime
388must be specified when using this option.
389.
390.TP
391.B \-\-test
392Do not actually change anything on the system, that is, the Clocks or
393.I @ADJTIME_PATH@
394.RB ( \%\-\-verbose
395is implicit with this option).
396.
397.TP
398.B \-\-update\-drift
399Update the Hardware Clock's drift factor in
400.IR @ADJTIME_PATH@ .
401It can only be used with
402.BR \-\-set " or " \%\-\-systohc ,
403.sp
404A minimum four hour period between settings is required. This is to
405avoid invalid calculations. The longer the period, the more precise the
406resulting drift factor will be.
407.sp
408This option was added in v2.26, because
409it is typical for systems to call
410.B \%hwclock\ \-\-systohc
411at shutdown; with the old behaviour this would automatically
412(re)calculate the drift factor which caused several problems:
413.RS
414.IP \(bu 2
415When using NTP with an \%'11\ minute\ mode' kernel the drift factor
416would be clobbered to near zero.
417.IP \(bu 2
418It would not allow the use of 'cold' drift correction. With most
419configurations using 'cold' drift will yield favorable results. Cold,
420means when the machine is turned off which can have a significant impact
421on the drift factor.
422.IP \(bu 2
423(Re)calculating drift factor on every shutdown delivers suboptimal
424results. For example, if ephemeral conditions cause the machine to be
425abnormally hot the drift factor calculation would be out of range.
426.IP \(bu 2
427Significantly increased system shutdown times (as of v2.31 when not
428using
429.B \%\-\-update\-drift
430the RTC is not read).
431.PP
432.RB "Having " \%hwclock
433calculate the drift factor is a good starting point, but for optimal
434results it will likely need to be adjusted by directly editing the
435.I @ADJTIME_PATH@
436file. For most configurations once a machine's optimal drift factor is
437crafted it should not need to be changed. Therefore, the old behavior to
438automatically (re)calculate drift was changed and now requires this
439option to be used. See the discussion below, under
440.BR "The Adjust Function" .
441.PP
442This option requires reading the Hardware Clock before setting it. If
443it cannot be read, then this option will cause the set functions to fail.
444This can happen, for example, if the Hardware Clock is corrupted by a
445power failure. In that case, the clock must first be set without this
446option. Despite it not working, the resulting drift correction factor
447would be invalid anyway.
448.RE
449.
450.TP
451.BR \-v ", " \-\-verbose
452Display more details about what
453.B \%hwclock
454is doing internally.
455.
456.SH NOTES
457.
458.SS Clocks in a Linux System
459.PP
460There are two types of date-time clocks:
461.PP
462.B The Hardware Clock:
463This clock is an independent hardware device, with its own power domain
464(battery, capacitor, etc), that operates when the machine is powered off,
465or even unplugged.
466.PP
467On an ISA compatible system, this clock is specified as part of the ISA
468standard. A control program can read or set this clock only to a whole
469second, but it can also detect the edges of the 1 second clock ticks, so
470the clock actually has virtually infinite precision.
471.PP
472This clock is commonly called the hardware clock, the real time clock,
473the RTC, the BIOS clock, and the CMOS clock. Hardware Clock, in its
474capitalized form, was coined for use by
475.BR \%hwclock .
476The Linux kernel also refers to it as the persistent clock.
477.PP
478Some non-ISA systems have a few real time clocks with
479only one of them having its own power domain.
480A very low power external I2C or SPI clock chip might be used with a
481backup battery as the hardware clock to initialize a more functional
482integrated real-time clock which is used for most other purposes.
483.PP
484.B The System Clock:
485This clock is part of the Linux kernel and is driven by
486a timer interrupt. (On an ISA machine, the timer interrupt is part of
487the ISA standard.) It has meaning only while Linux is running on the
488machine. The System Time is the number of seconds since 00:00:00
489January 1, 1970 UTC (or more succinctly, the number of seconds since
4901969 UTC). The System Time is not an integer, though. It has virtually
491infinite precision.
492.PP
493The System Time is the time that matters. The Hardware Clock's basic
494purpose is to keep time when Linux is not running so that the System
495Clock can be initialized from it at boot. Note that in DOS, for which
496ISA was designed, the Hardware Clock is the only real time clock.
497.PP
498It is important that the System Time not have any discontinuities such as
499would happen if you used the
500.BR \%date (1)
501program to set it while the system is running. You can, however, do whatever
502you want to the Hardware Clock while the system is running, and the next
503time Linux starts up, it will do so with the adjusted time from the Hardware
504Clock. Note: currently this is not possible on most systems because
505.B \%hwclock\ \-\-systohc
506is called at shutdown.
507.PP
508The Linux kernel's timezone is set by
509.BR hwclock .
510But don't be misled -- almost nobody cares what timezone the kernel
511thinks it is in. Instead, programs that care about the timezone
512(perhaps because they want to display a local time for you) almost
513always use a more traditional method of determining the timezone: They
514use the TZ environment variable or the
515.I \%/etc/localtime
516file, as explained in the man page for
517.BR \%tzset (3).
518However, some programs and fringe parts of the Linux kernel such as filesystems
519use the kernel's timezone value. An example is the vfat filesystem. If the
520kernel timezone value is wrong, the vfat filesystem will report and set the
521wrong timestamps on files. Another example is the kernel's NTP \%'11\ minute\ mode'.
522If the kernel's timezone value and/or the
523.I \%persistent_clock_is_local
524variable are wrong, then the Hardware Clock will be set incorrectly
525by \%'11\ minute\ mode'. See the discussion below, under
526.BR "Automatic Hardware Clock Synchronization by the Kernel" .
527.PP
528.B \%hwclock
529sets the kernel's timezone to the value indicated by TZ or
530.IR \%/etc/localtime " with the"
531.BR \%\-\-hctosys " or " \%\-\-systz " functions."
532.PP
533The kernel's timezone value actually consists of two parts: 1) a field
534tz_minuteswest indicating how many minutes local time (not adjusted
535for DST) lags behind UTC, and 2) a field tz_dsttime indicating
536the type of Daylight Savings Time (DST) convention that is in effect
537in the locality at the present time.
538This second field is not used under Linux and is always zero.
539See also
540.BR \%settimeofday (2).
541.
542.SS Hardware Clock Access Methods
543.PP
544.B \%hwclock
545uses many different ways to get and set Hardware Clock values. The most
546normal way is to do I/O to the rtc device special file, which is
547presumed to be driven by the rtc device driver. Also, Linux systems
548using the rtc framework with udev, are capable of supporting multiple
549Hardware Clocks. This may bring about the need to override the default
550rtc device by specifying one with the
551.BR \-\-rtc " option."
552.PP
553However, this method is not always available as older systems do not
554have an rtc driver. On these systems, the method of accessing the
555Hardware Clock depends on the system hardware.
556.PP
557On an ISA compatible system,
558.B \%hwclock
559can directly access the "CMOS memory" registers that
560constitute the clock, by doing I/O to Ports 0x70 and 0x71. It does
561this with actual I/O instructions and consequently can only do it if
562running with superuser effective userid. This method may be used by
563specifying the
564.BR \%\-\-directisa " option."
565.PP
566This is a really poor method of accessing the clock, for all the
567reasons that userspace programs are generally not supposed to do
568direct I/O and disable interrupts.
569.B \%hwclock
570provides it for testing, troubleshooting, and because it may be the
571only method available on ISA systems which do not have a working rtc
572device driver.
573.SS The Adjust Function
574.PP
575The Hardware Clock is usually not very accurate. However, much of its
576inaccuracy is completely predictable - it gains or loses the same amount
577of time every day. This is called systematic drift.
578.BR \%hwclock "'s " \%\-\-adjust
579function lets you apply systematic drift corrections to the
580Hardware Clock.
581.PP
582It works like this:
583.BR \%hwclock " keeps a file,"
584.IR @ADJTIME_PATH@ ,
585that keeps some historical information. This is called the adjtime file.
586.PP
587Suppose you start with no adjtime file. You issue a
588.B \%hwclock\ \-\-set
589command to set the Hardware Clock to the true current time.
590.B \%hwclock
591creates the adjtime file and records in it the current time as the
592last time the clock was calibrated.
593Five days later, the clock has gained 10 seconds, so you issue a
594.B \%hwclock\ \-\-set\ \-\-update\-drift
595command to set it back 10 seconds.
596.B \%hwclock
597updates the adjtime file to show the current time as the last time the
598clock was calibrated, and records 2 seconds per day as the systematic
599drift rate. 24 hours go by, and then you issue a
600.B \%hwclock\ \-\-adjust
601command.
602.B \%hwclock
603consults the adjtime file and sees that the clock gains 2 seconds per
604day when left alone and that it has been left alone for exactly one
605day. So it subtracts 2 seconds from the Hardware Clock. It then
606records the current time as the last time the clock was adjusted.
607Another 24 hours go by and you issue another
608.BR \%hwclock\ \-\-adjust .
609.B \%hwclock
610does the same thing: subtracts 2 seconds and updates the adjtime file
611with the current time as the last time the clock was adjusted.
612.PP
613When you use the
614.BR \%\-\-update\-drift " option with " \-\-set " or " \%\-\-systohc ,
615the systematic drift rate is (re)calculated by comparing the fully drift
616corrected current Hardware Clock time with the new set time, from that
617it derives the 24 hour drift rate based on the last calibrated timestamp
618from the adjtime file. This updated drift factor is then saved in
619.IR @ADJTIME_PATH@ .
620.PP
621A small amount of error creeps in when
622the Hardware Clock is set, so
623.B \%\-\-adjust
624refrains from making any adjustment that is less
625than 1 second. Later on, when you request an adjustment again, the accumulated
626drift will be more than 1 second and
627.B \%\-\-adjust
628will make the adjustment including any fractional amount.
629.PP
630.B \%hwclock\ \-\-hctosys
631also uses the adjtime file data to compensate the value read from the Hardware
632Clock before using it to set the System Clock. It does not share the 1 second
633limitation of
634.BR \%\-\-adjust ,
635and will correct sub-second drift values immediately. It does not
636change the Hardware Clock time nor the adjtime file. This may eliminate
637the need to use
638.BR \%\-\-adjust ,
639unless something else on the system needs the Hardware Clock to be
640compensated.
641.
642.SS The Adjtime File
643While named for its historical purpose of controlling adjustments only,
644it actually contains other information used by
645.B hwclock
646from one invocation to the next.
647.PP
648The format of the adjtime file is, in ASCII:
649.PP
650Line 1: Three numbers, separated by blanks: 1) the systematic drift rate
651in seconds per day, floating point decimal; 2) the resulting number of
652seconds since 1969 UTC of most recent adjustment or calibration,
653decimal integer; 3) zero (for compatibility with
654.BR \%clock (8))
655as a decimal integer.
656.PP
657Line 2: One number: the resulting number of seconds since 1969 UTC of most
658recent calibration. Zero if there has been no calibration yet or it
659is known that any previous calibration is moot (for example, because
660the Hardware Clock has been found, since that calibration, not to
661contain a valid time). This is a decimal integer.
662.PP
663Line 3: "UTC" or "LOCAL". Tells whether the Hardware Clock is set to
664Coordinated Universal Time or local time. You can always override this
665value with options on the
666.B \%hwclock
667command line.
668.PP
669You can use an adjtime file that was previously used with the
670.BR \%clock "(8) program with " \%hwclock .
671.
672.SS Automatic Hardware Clock Synchronization by the Kernel
673.PP
674You should be aware of another way that the Hardware Clock is kept
675synchronized in some systems. The Linux kernel has a mode wherein it
676copies the System Time to the Hardware Clock every 11 minutes. This mode
677is a compile time option, so not all kernels will have this capability.
678This is a good mode to use when you are using something sophisticated
679like NTP to keep your System Clock synchronized. (NTP is a way to keep
680your System Time synchronized either to a time server somewhere on the
681network or to a radio clock hooked up to your system. See RFC 1305.)
682.PP
683If the kernel is compiled with the \%'11\ minute\ mode' option it will
684be active when the kernel's clock discipline is in a synchronized state.
685When in this state, bit 6 (the bit that is set in the mask 0x0040)
686of the kernel's
687.I \%time_status
688variable is unset. This value is output as the 'status' line of the
689.BR \%adjtimex\ --print " or " \%ntptime " commands."
690.PP
691It takes an outside influence, like the NTP daemon
692to put the kernel's clock discipline into a synchronized state, and
693therefore turn on \%'11\ minute\ mode'.
694It can be turned off by running anything that sets the System Clock the old
695fashioned way, including
696.BR "\%hwclock\ \-\-hctosys" .
697However, if the NTP daemon is still running, it will turn \%'11\ minute\ mode'
698back on again the next time it synchronizes the System Clock.
699.PP
700If your system runs with \%'11\ minute\ mode' on, it may need to use either
701.BR \%\-\-hctosys " or " \%\-\-systz
702in a startup script, especially if the Hardware Clock is configured to use
703the local timescale. Unless the kernel is informed of what timescale the
704Hardware Clock is using, it may clobber it with the wrong one. The kernel
705uses UTC by default.
706.PP
707The first userspace command to set the System Clock informs the
708kernel what timescale the Hardware Clock is using. This happens via the
709.I \%persistent_clock_is_local
710kernel variable. If
711.BR \%\-\-hctosys " or " \%\-\-systz
712is the first, it will set this variable according to the adjtime file or the
713appropriate command-line argument. Note that when using this capability and the
714Hardware Clock timescale configuration is changed, then a reboot is required to
715notify the kernel.
716.PP
717.B \%hwclock\ \-\-adjust
718should not be used with NTP \%'11\ minute\ mode'.
719.
720.SS ISA Hardware Clock Century value
721.PP
722There is some sort of standard that defines CMOS memory Byte 50 on an ISA
723machine as an indicator of what century it is.
724.B \%hwclock
725does not use or set that byte because there are some machines that
726don't define the byte that way, and it really isn't necessary anyway,
727since the year-of-century does a good job of implying which century it
728is.
729.PP
730If you have a bona fide use for a CMOS century byte, contact the
731.B \%hwclock
732maintainer; an option may be appropriate.
733.PP
734Note that this section is only relevant when you are using the "direct
735ISA" method of accessing the Hardware Clock.
736ACPI provides a standard way to access century values, when they
737are supported by the hardware.
738.
739.SH DATE-TIME CONFIGURATION
740.in +4
741.SS Keeping Time without External Synchronization
742.in
743.PP
744This discussion is based on the following conditions:
745.IP \(bu 2
746Nothing is running that alters the date-time clocks, such as NTP daemon or a cron job."
747.IP \(bu 2
748The system timezone is configured for the correct local time. See below, under
749.BR "POSIX vs 'RIGHT'" .
750.IP \(bu 2
751Early during startup the following are called, in this order:
752.br
753.BI \%adjtimex\ \-\-tick \ value\ \-\-frequency \ value
754.br
755.B \%hwclock\ \-\-hctosys
756.IP \(bu 2
757During shutdown the following is called:
758.br
759.B \%hwclock\ \-\-systohc
760.PP
761.in +4
762.BR * " Systems without " adjtimex " may use " ntptime .
763.in
764.PP
765Whether maintaining precision time with NTP daemon
766or not, it makes sense to configure the system to keep reasonably good
767date-time on its own.
768.PP
769The first step in making that happen is having a clear understanding of
770the big picture. There are two completely separate hardware devices
771running at their own speed and drifting away from the 'correct' time at
772their own rates. The methods and software for drift correction are
773different for each of them. However, most systems are configured to
774exchange values between these two clocks at startup and shutdown. Now
775the individual device's time keeping errors are transferred back and
776forth between each other. Attempt to configure drift correction for only
777one of them, and the other's drift will be overlaid upon it.
778.PP
779This problem can be avoided when configuring drift correction for the
780System Clock by simply not shutting down the machine. This, plus the
781fact that all of
782.BR \%hwclock 's
783precision (including calculating drift factors) depends upon the System
784Clock's rate being correct, means that configuration of the System Clock
785should be done first.
786.PP
787The System Clock drift is corrected with the
788.BR \%adjtimex "(8) command's " \-\-tick " and " \%\-\-frequency
789options. These two work together: tick is the coarse adjustment and
790frequency is the fine adjustment. (For systems that do not have an
791.BR \%adjtimex " package,"
792.BI \%ntptime\ \-f\ ppm
793may be used instead.)
794.PP
795Some Linux distributions attempt to automatically calculate the System
796Clock drift with
797.BR \%adjtimex 's
798compare operation. Trying to correct one
799drifting clock by using another drifting clock as a reference is akin to
800a dog trying to catch its own tail. Success may happen eventually, but
801great effort and frustration will likely precede it. This automation may
802yield an improvement over no configuration, but expecting optimum
803results would be in error. A better choice for manual configuration
804would be
805.BR \%adjtimex 's " \-\-log " options.
806.PP
807It may be more effective to simply track the System Clock drift with
808.BR \%sntp ", or " \%date\ \-Ins
809and a precision timepiece, and then calculate the correction manually.
810.PP
811After setting the tick and frequency values, continue to test and refine the
812adjustments until the System Clock keeps good time. See
813.BR \%adjtimex (8)
814for more information and the example demonstrating manual drift
815calculations.
816.PP
817Once the System Clock is ticking smoothly, move on to the Hardware Clock.
818.PP
819As a rule, cold drift will work best for most use cases. This should be
820true even for 24/7 machines whose normal downtime consists of a reboot.
821In that case the drift factor value makes little difference. But on the
822rare occasion that the machine is shut down for an extended period, then
823cold drift should yield better results.
824.PP
825.B Steps to calculate cold drift:
826.IP 1 2
827.B "Ensure that NTP daemon will not be launched at startup."
828.IP 2 2
829.RI The " System Clock " "time must be correct at shutdown!"
830.IP 3 2
831Shut down the system.
832.IP 4 2
833Let an extended period pass without changing the Hardware Clock.
834.IP 5 2
835Start the system.
836.IP 6 2
837.RB "Immediately use " hwclock " to set the correct time, adding the"
838.BR \%\-\-update\-drift " option."
839.PP
840Note: if step 6 uses
841.BR \%\-\-systohc ,
842then the System Clock must be set correctly (step 6a) just before doing so.
843.PP
844.RB "Having " hwclock
845calculate the drift factor is a good starting point, but for optimal
846results it will likely need to be adjusted by directly editing the
847.I @ADJTIME_PATH@
848file. Continue to test and refine the drift factor until the Hardware
849Clock is corrected properly at startup. To check this, first make sure
850that the System Time is correct before shutdown and then use
851.BR \%sntp ", or " \%date\ \-Ins
852and a precision timepiece, immediately after startup.
853.SS LOCAL vs UTC
854Keeping the Hardware Clock in a local timescale causes inconsistent
855daylight saving time results:
856.IP \(bu 2
857If Linux is running during a daylight saving time change, the time
858written to the Hardware Clock will be adjusted for the change.
859.IP \(bu 2
860If Linux is NOT running during a daylight saving time change, the time
861read from the Hardware Clock will NOT be adjusted for the change.
862.PP
863The Hardware Clock on an ISA compatible system keeps only a date and time,
864it has no concept of timezone nor daylight saving. Therefore, when
865.B hwclock
866is told that it is in local time, it assumes it is in the 'correct'
867local time and makes no adjustments to the time read from it.
868.PP
869Linux handles daylight saving time changes transparently only when the
870Hardware Clock is kept in the UTC timescale. Doing so is made easy for
871system administrators as
872.B \%hwclock
873uses local time for its output and as the argument to the
874.BR \%\-\-date " option."
875.PP
876POSIX systems, like Linux, are designed to have the System Clock operate
877in the UTC timescale. The Hardware Clock's purpose is to initialize the
878System Clock, so also keeping it in UTC makes sense.
879.PP
880Linux does, however, attempt to accommodate the Hardware Clock being in
881the local timescale. This is primarily for dual-booting with older
882versions of MS Windows. From Windows 7 on, the RealTimeIsUniversal
883registry key is supposed to be working properly so that its Hardware
884Clock can be kept in UTC.
885.
886.SS POSIX vs 'RIGHT'
887A discussion on date-time configuration would be incomplete without
888addressing timezones, this is mostly well covered by
889.BR tzset (3).
890One area that seems to have no documentation is the 'right'
891directory of the Time Zone Database, sometimes called tz or zoneinfo.
892.PP
893There are two separate databases in the zoneinfo system, posix
894and 'right'. 'Right' (now named zoneinfo\-leaps) includes leap seconds and posix
895does not. To use the 'right' database the System Clock must be set to
896\%(UTC\ +\ leap seconds), which is equivalent to \%(TAI\ \-\ 10). This
897allows calculating the
898exact number of seconds between two dates that cross a leap second
899epoch. The System Clock is then converted to the correct civil time,
900including UTC, by using the 'right' timezone files which subtract the
901leap seconds. Note: this configuration is considered experimental and is
902known to have issues.
903.PP
904To configure a system to use a particular database all of the files
905located in its directory must be copied to the root of
906.IR \%/usr/share/zoneinfo .
907Files are never used directly from the posix or 'right' subdirectories, e.g.,
908.RI \%TZ=' right/Europe/Dublin '.
909This habit was becoming so common that the upstream zoneinfo project
910restructured the system's file tree by moving the posix and 'right'
911subdirectories out of the zoneinfo directory and into sibling directories:
912.PP
913.in +2
914.I /usr/share/zoneinfo
915.br
916.I /usr/share/zoneinfo\-posix
917.br
918.I /usr/share/zoneinfo\-leaps
919.PP
920Unfortunately, some Linux distributions are changing it back to the old
921tree structure in their packages. So the problem of system
922administrators reaching into the 'right' subdirectory persists. This
923causes the system timezone to be configured to include leap seconds
924while the zoneinfo database is still configured to exclude them. Then
925when an application such as a World Clock needs the South_Pole timezone
926file; or an email MTA, or
927.B hwclock
928needs the UTC timezone file; they fetch it from the root of
929.I \%/usr/share/zoneinfo
930, because that is what they are supposed to do. Those files exclude leap
931seconds, but the System Clock now includes them, causing an incorrect
932time conversion.
933.PP
934Attempting to mix and match files from these separate databases will not
935work, because they each require the System Clock to use a different
936timescale. The zoneinfo database must be configured to use either posix
937or 'right', as described above, or by assigning a database path to the
938.SB TZDIR
939environment variable.
940.SH EXIT STATUS
941One of the following exit values will be returned:
942.TP
943.BR EXIT_SUCCESS " ('0' on POSIX systems)"
944Successful program execution.
945.TP
946.BR EXIT_FAILURE " ('1' on POSIX systems)"
947The operation failed or the command syntax was not valid.
948.SH ENVIRONMENT
949.TP
950.B TZ
951If this variable is set its value takes precedence over the system
952configured timezone.
953.TP
954.B TZDIR
955If this variable is set its value takes precedence over the system
956configured timezone database directory path.
957.SH FILES
958.TP
959.I @ADJTIME_PATH@
960The configuration and state file for hwclock.
961.TP
962.I /etc/localtime
963The system timezone file.
964.TP
965.I /usr/share/zoneinfo/
966The system timezone database directory.
967.PP
968Device files
969.B hwclock
970may try for Hardware Clock access:
971.br
972.I /dev/rtc0
973.br
974.I /dev/rtc
975.br
976.I /dev/misc/rtc
977.br
978.I /dev/efirtc
979.br
980.I /dev/misc/efirtc
981.SH "SEE ALSO"
982.BR date (1),
983.BR adjtimex (8),
984.BR gettimeofday (2),
985.BR settimeofday (2),
986.BR crontab (1),
987.BR tzset (3)
988.
989.SH AUTHORS
990Written by Bryan Henderson, September 1996 (bryanh@giraffe-data.com),
991based on work done on the
992.BR \%clock (8)
993program by Charles Hedrick, Rob Hooft, and Harald Koenig.
994See the source code for complete history and credits.
995.
996.SH AVAILABILITY
997The hwclock command is part of the util-linux package and is available from
998https://www.kernel.org/pub/linux/utils/util-linux/.