Clock drift on an iBook2

Clock drift on an iBook2

[Ethan Blanton]

[-- Attachment #1: Type: text/plain, Size: 512 bytes --]

Hi,
I have a new iBook2 and I'm having *severe* clock drift problems -- on
the order of several seconds gained per minute.  I'd like to get this
taken care of, and it seems to be a linux problem as OSX appears to
have a stable clock.

I'm running BenH's 2.4.7-pre6 right now, but I also saw it on 2.4.6.

Is this a known problem?  If not, what can I do to help get this fixed?
Ethan

-- 
If I've told you once, I've told you once
And once is all that you needed.
		-- The Refreshments, "Carefree"

[-- Attachment #2: Type: application/pgp-signature, Size: 0 bytes --]

Re: Clock drift on an iBook2

[Benjamin Herrenschmidt]

>I have a new iBook2 and I'm having *severe* clock drift problems -- on
>the order of several seconds gained per minute.  I'd like to get this
>taken care of, and it seems to be a linux problem as OSX appears to
>have a stable clock.
>
>I'm running BenH's 2.4.7-pre6 right now, but I also saw it on 2.4.6.
>
>Is this a known problem?  If not, what can I do to help get this fixed?
>Ethan

Are you using MOL or do you experience this with nothing special running ?

Do you have the same issue if you disable powersave_nap (/proc/sys/
kernel/powersave_nap) ?

The TB is calibrated using the VIA, if we don't have a correct VIA frequency
or if the calibration routine is slightly off, that may also explain the
problem.

Ben.


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: Clock drift on an iBook2

[Benjamin Herrenschmidt]

>>I have a new iBook2 and I'm having *severe* clock drift problems -- on
>>the order of several seconds gained per minute.  I'd like to get this
>>taken care of, and it seems to be a linux problem as OSX appears to
>>have a stable clock.
>>
>>I'm running BenH's 2.4.7-pre6 right now, but I also saw it on 2.4.6.
>>
>>Is this a known problem?  If not, what can I do to help get this fixed?
>>Ethan
>
>Are you using MOL or do you experience this with nothing special running ?
>
>Do you have the same issue if you disable powersave_nap (/proc/sys/
>kernel/powersave_nap) ?
>
>The TB is calibrated using the VIA, if we don't have a correct VIA frequency
>or if the calibration routine is slightly off, that may also explain the
>problem.

Something else you can try: in arch/ppc/kernel/pmac_time.c, comment
out the call to via_calibrate_decr() (and the return too) so that
we use the device-tree "timebase-frequency" instead, and tell us if
it makes any difference.

Ben.


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: Clock drift on an iBook2

[Ethan Blanton]

[-- Attachment #1: Type: text/plain, Size: 1092 bytes --]

Benjamin Herrenschmidt spake unto us the following wisdom:
> >I have a new iBook2 and I'm having *severe* clock drift problems -- on
> >the order of several seconds gained per minute.  I'd like to get this
> >taken care of, and it seems to be a linux problem as OSX appears to
> >have a stable clock.
> 
> Are you using MOL or do you experience this with nothing special running ?

Nothing special...   I tried using ntpd to fix the problem, but it
didn't seem to be able to set the system time and I'm not sure why.

> Do you have the same issue if you disable powersave_nap (/proc/sys/
> kernel/powersave_nap) ?

Yes, it seems to make no difference.  (~4s of drift in 3 minutes
either way, according to the tests I just did)

> The TB is calibrated using the VIA, if we don't have a correct VIA frequency
> or if the calibration routine is slightly off, that may also explain the
> problem.

Looking into your next email regarding the VIA...
Ethan

-- 
If I've told you once, I've told you once
And once is all that you needed.
		-- The Refreshments, "Carefree"

[-- Attachment #2: Type: application/pgp-signature, Size: 0 bytes --]

Re: Clock drift on an iBook2

[Tom Rini]

On Tue, Jul 17, 2001 at 01:04:25PM -0400, Ethan Blanton wrote:
> Benjamin Herrenschmidt spake unto us the following wisdom:
> > >I have a new iBook2 and I'm having *severe* clock drift problems -- on
> > >the order of several seconds gained per minute.  I'd like to get this
> > >taken care of, and it seems to be a linux problem as OSX appears to
> > >have a stable clock.
> >
> > Are you using MOL or do you experience this with nothing special running ?
>
> Nothing special...   I tried using ntpd to fix the problem, but it
> didn't seem to be able to set the system time and I'm not sure why.

Does your kernel have CONFIG_PPC_RTC enabled? (_not_ CONFIG_RTC).

--
Tom Rini (TR1265)
http://gate.crashing.org/~trini/

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: Clock drift on an iBook2

[Ethan Blanton]

[-- Attachment #1: Type: text/plain, Size: 420 bytes --]

Tom Rini spake unto us the following wisdom:
> > Nothing special...   I tried using ntpd to fix the problem, but it
> > didn't seem to be able to set the system time and I'm not sure why.
> 
> Does your kernel have CONFIG_PPC_RTC enabled? (_not_ CONFIG_RTC).

Yes ... and ntpdate works.
Ethan

-- 
If I've told you once, I've told you once
And once is all that you needed.
		-- The Refreshments, "Carefree"

[-- Attachment #2: Type: application/pgp-signature, Size: 0 bytes --]

Re: Clock drift on an iBook2

[Ethan Blanton]

[-- Attachment #1: Type: text/plain, Size: 667 bytes --]

Benjamin Herrenschmidt spake unto us the following wisdom:
> Something else you can try: in arch/ppc/kernel/pmac_time.c, comment
> out the call to via_calibrate_decr() (and the return too) so that
> we use the device-tree "timebase-frequency" instead, and tell us if
> it makes any difference.

That seems to have improved it *greatly*.   I know have a drift (only
sampled after a few minutes, so somewhat inaccurate -- but accurate
enough to tell that it's no longer seconds/minute) more on a par with
what I have come to expect.
Ethan

-- 
If I've told you once, I've told you once
And once is all that you needed.
		-- The Refreshments, "Carefree"

[-- Attachment #2: Type: application/pgp-signature, Size: 0 bytes --]