1 |
William Kenworthy <billk@×××××××××.au> [13-08-25 06:32]: |
2 |
> or if you have an initramfs do it there before control is passed to the |
3 |
> OS. You may also be able to set rtc0 to the current time in the |
4 |
> initramfs as well. |
5 |
> |
6 |
> BillK |
7 |
> |
8 |
> |
9 |
> |
10 |
> On 25/08/13 12:18, Mark David Dumlao wrote: |
11 |
> > Which runlevel did you put your script on? |
12 |
> > You probably want it on sysinit, rather than default. |
13 |
> > |
14 |
> > Also, you can put rc_before= and rc_after= in the corresponding |
15 |
> > /etc/conf.d/ file and make sure it runs before your syslog but after |
16 |
> > your sysfs. |
17 |
> > |
18 |
> > On Sun, Aug 25, 2013 at 12:04 PM, <meino.cramer@×××.de> wrote: |
19 |
> >> Hi, |
20 |
> >> |
21 |
> >> There are two RTCs in my system: |
22 |
> >> /dev/rtc0 and /dev/rtc1 |
23 |
> >> |
24 |
> >> rtc0 is not powered by a battery and forgets time/date |
25 |
> >> with system shutdown and rtc1 is a I2C-rtc (DS3231) which |
26 |
> >> is powered by a battery. It is extremly accurate in comparison |
27 |
> >> with rtc0. |
28 |
> >> |
29 |
> >> rtc0 is accessible with system boot - rtc1 is not (current state). |
30 |
> >> |
31 |
> >> To make rtc1 completly know to the system, I have to do a |
32 |
> >> echo ds3231 0x68 >! /sys/bus/i2c/devices/i2c-1/new_device |
33 |
> >> hwclock -f /dev/rtc1 -s |
34 |
> >> in beforehand. |
35 |
> >> |
36 |
> >> I wrote a script for /etc/init.d, which does exactly this, and |
37 |
> >> the system login screen shows the correct tome/date information |
38 |
> >> even without ntp-client (the system should become independantly |
39 |
> >> from lan/internet). The script was added to the boot service |
40 |
> >> and executed after dev and modules. |
41 |
> >> |
42 |
> >> The kernel is configured to use /dev/rtc1 and the driver for the |
43 |
> >> ds1307, which also handles the ds3231, is included into the kernel |
44 |
> >> (no module). |
45 |
> >> |
46 |
> >> But it seems, that setting the system time this way is too late, |
47 |
> >> since there are still (for example) log files under /var/log |
48 |
> >> with a timestamp of the 1.1.1970. |
49 |
> >> |
50 |
> >> Are there any other way to make rtc1 known and accessible earlier |
51 |
> >> to the system as the hack via a script in /etc/init.d ? |
52 |
> >> |
53 |
> >> Thank you very much in advance for any help! |
54 |
> >> Best regards, |
55 |
> >> mcc |
56 |
> >> |
57 |
> >> |
58 |
> >> |
59 |
> >> |
60 |
> > |
61 |
> > |
62 |
> > |
63 |
> |
64 |
> |
65 |
|
66 |
Hi Mark, hi William, |
67 |
|
68 |
the script ds3231 in /etc/init.d is -- according to rc-update -- |
69 |
set as folows: |
70 |
|
71 |
ds3231 | boot |
72 |
|
73 |
There is no corresponding file in /etc/conf.d since the script |
74 |
onlu consist of two commands (see previous posting). There is no |
75 |
initramfs. |
76 |
|
77 |
Since hwclock needs /dev/rtc1 to work, the script cannot be executed |
78 |
before /dev/. is up. |
79 |
|
80 |
Is there an text/tutorial/... or such somewhere on the net, which |
81 |
shows the sequence, in which the usual services/scripts in /etc/init.d |
82 |
get started/executed while the system boots? |
83 |
|
84 |
Best regads, |
85 |
mcc |