System time not staying accurate
 

I'm running RH9 on a Dell Latitude c840 1MB Ram 60GB HD P4 1.6 Mhz

I am running a dual boot ... when ever I run under RH9 my system time doesn't stay current. Even when I come back up under windoze the time is incorrect. I did a quick search but didn't see anything that sounded right.

Any ideas?

Thanks in advance

Your system clock is faulty; this is common enough.

See the Managing Accurate Date and Time How-To from the Linux Documentation Project.

That doesn't sit right ... I'lll fully read that doc in just a sec but at first glance it's just talking about setting the clock ... When under windoze the time is kept correctly ... but under RH9 ... the clock runs slow ... when I go back to win from RH9 the clock is incorrect. .... I'll fine tooth comb that doc and see if anything makes sense

the doc gives me a work around to the problem but doesn't explain why the system clock runs slow under RH9 ... my guess is that I made a wrong choice when installing, telling RH9 that I had a different proc then I really do so the system forces the proc to run at .9 instead of 1 (for example)


Possible? or am I way off?

If the clock drifts while the computer is off, this is a problem with the CMOS clock. If the clock drifts while you are booted into Linux, this is a problem with your kernel configuration.

One possible reason the windows clock is accuracte and the Linux clock is not could be that Windows is automatically syncing your clock to the internet and Linux is not. To sync your Linux clock to the internet, run: "ntpdate time.nist.gov" every time the clock starts to drift. Or, install ntpd, which will automatically detect when your clock drifts and resync it for you.

The reason the Linux time appears different after using Windows may be that Windows stores the time based on your local time zone and Linux stores it in UTM. When Windows modifies the time to your local time zone and Linux tries to read in in UTM, you see a different time. This would be further evidence that Windows is syncing your clock to the internet to compensate for drift.

I'm not sure what you mean by forcing proc to run at different speeds. I was unaware proc had a speed, let alone that you could change it :confused:.

We were guessing the processor was being "clocked down" ...just windoze users guessing at what might be happening.

The clock is correct when I first boot back into Linux (if I correct it in windoze) I'm not syncing the clock with any outside source. If it is incorrect it stays incorrect until I change it. windoze or otherwise.

I'm guessing if the there was a CMOS problem the drifting would happen under windoze too. THis only happens when booted into Linux. I am interested to see if the time drifts when I boot into Linux ... shut down for 24 hours and then boot again into linux

I know it is not a syncing/time zone problem ... I can use the syncing to fix the clock displayed time problem but me and mine are guessing that since the clock is running slow that this is the symptom of a larger problem. I have heard of over clocking the processor (causing more "thinking" cycles and thus gaing a speed boost on a processor ... a common side effect of doing that would be that ones clock would run fast ... so we are guessing what I'm experiencing is that some how I've set my system up to run at say (for example) 1.6 GHz instead of 1.8Ghz ... the cmos is expecting the system to be running at 1.8 and thus counts the thinking cycles to judge time. Normally the proc runs 1.8 in one second but since it is (I'm guessing) running at 1.6 it takes the 1.6 (in one second) + the next .2 cycles before the system says (Hey a second has passed increase the clock count by one!)

As kc8tbe stated it is most likely some kind of a kernel config problem ... Where can I go to check what the kernel thinks it should be running at? can I change this without formatting and re-installing?
:scratch:

You are correct, what you are describing is definitely a kernel config problem.

I'm fairly certain that processor clock speed is only a problem if:
1. You are using a Motherboard/processor that automatically clocks down the processor to save power (i.e. a laptop motherboard).
2. You compiled processor cycling into the kernel.
3. You changed the bus speeds on your motherboard so as to overclock your proccessor. You would have to manually open of the computer and play with it to do this.

Of course, your problem can be in the kernel config without your problem being related to clocking. Do you specify the correct processor type in the kernel config?

I don't know the command to see what your processor speed is, but on bootup a "BogoMips" number will be displayed. Compare that number to the table found here:
http://docs.mandragor.org/files/Oper...s_en/misc.html

It is a laptop motherboard ... being the newbie I am I pretty much just took all the defaults that the gui install threw at me ... I'll compare the numbers in a bit and post back ... thanks

Here is what I found in the syslog (I haven't yet found how to turn the BogoMIPS into a value on that table)

Jan 2 16:21:15 localhost kernel: Initializing CPU#0
Jan 2 16:21:15 localhost irqbalance: irqbalance startup succeeded
Jan 2 16:21:15 localhost kernel: Detected 1196.488 MHz processor.
Jan 2 16:21:15 localhost kernel: Console: colour VGA+ 80x25
Jan 2 16:21:15 localhost kernel: Calibrating delay loop... 2385.51 BogoMIPS
Jan 2 16:21:15 localhost kernel: Memory: 1027340k/1048456k available (1359k kernel code, 17532k reserved, 1004k data, 132k init, 130952k highmem)
Jan 2 16:21:15 localhost kernel: Dentry cache hash table entries: 131072 (order: 8, 1048576 bytes)
Jan 2 16:21:15 localhost portmap: portmap startup succeeded
Jan 2 16:21:15 localhost kernel: Inode cache hash table entries: 65536 (order: 7, 524288 bytes)
Jan 2 16:21:15 localhost kernel: Mount cache hash table entries: 512 (order: 0, 4096 bytes)
Jan 2 16:21:15 localhost kernel: Buffer-cache hash table entries: 65536 (order: 6, 262144 bytes)
Jan 2 16:21:15 localhost kernel: Page-cache hash table entries: 262144 (order: 8, 1048576 bytes)
Jan 2 16:21:15 localhost kernel: CPU: Trace cache: 12K uops, L1 D cache: 8K
Jan 2 16:21:11 localhost network: Setting network parameters: succeeded
Jan 2 16:21:16 localhost kernel: CPU: L2 cache: 512K
Jan 2 16:21:16 localhost rpc.statd[2228]: Version 1.0.1 Starting
Jan 2 16:21:16 localhost nfslock: rpc.statd startup succeeded
Jan 2 16:21:11 localhost logger: punching nameserver XXXXXXX through the firewall
Jan 2 16:21:16 localhost kernel: Intel machine check architecture supported.
Jan 2 16:21:11 localhost logger: punching nameserver XXXXXXX through the firewall
Jan 2 16:21:16 localhost keytable: Loading keymap:
Jan 2 16:21:16 localhost kernel: Intel machine check reporting enabled on CPU#0.
Jan 2 16:21:11 localhost network: Bringing up loopback interface: succeeded
Jan 2 16:21:11 localhost ifup:
Jan 2 16:21:11 localhost ifup: Determining IP information for eth0...
Jan 2 16:21:16 localhost kernel: CPU: Intel(R) Pentium(R) 4 Mobile CPU 1.60GHz stepping 04

After reading the tables and finally understanding them it looks like my MIPs should be somewhere in the realm of 3200 but as we see from the boot log I'm getting 2800 .... so it appears there is something going on to slow the system down ...

now where? (THanks again for everyones help and time)

I'm ready to re-install as a BIG HAMMER APPROACH but would prefer not to (plus I don't know if that would really change anything cause all I did was push the next button during the original install)

> Jan 2 16:21:15 localhost kernel: Detected 1196.488 MHz processor
> Jan 2 16:21:16 localhost kernel: CPU: Intel(R) Pentium(R) 4 Mobile CPU 1.60GHz stepping

These tell me two things:
1. Regardless of what your CPU speed actually is, your system can't seem to decide. First your clock speed is 1196.488 MHz, then it is 1.6 GHz. This could be causing a clock speed problem.
2. "stepping" is mentioned, which means there's a good chance that something is throttling the CPU to use battery power more efficiently. This could also cause system clock woes.

Unfortunately, your problem is now beyond what I can convincingly b.s. :( Somebody with more knowledge and expertise will be able to help you.
My only other suggestion would be to compile Dell laptop support into the kernel (something you've probably already done), or try a different kernel version. Best of luck getting your clock to work!