frequency in sandpoint_calibrate_decr

frequency in sandpoint_calibrate_decr

[Sébastien Côté]

Hi,

In MontaVista's kernel 2.4.0-test2, the frequency in
sandpoint_calibrate_decr() is set to 66 Mhz.  What frequency is this?
The processor, the bus ???

Increasing this speed seems to get rid of my stability problems (at
last!), so I'd like to know the real value I should put there.

Thanks,

Sébas..

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

Re: frequency in sandpoint_calibrate_decr

[Dan Malek]

Sébastien Côté wrote:

> In MontaVista's kernel 2.4.0-test2, the frequency in
> sandpoint_calibrate_decr() is set to 66 Mhz.  What frequency is this?
> The processor, the bus ???

That's the speed of the processor bus, usually from which all
other core clocks are derived.  You should be using a newer version
of the kernel, at least from FSM Labs BitKeeper.  In the source
code pipeline is a change of the clock value from MHz to Hz.

> Increasing this speed seems to get rid of my stability problems (at
> last!), so I'd like to know the real value I should put there.

Eh?  On most systems it won't do anything but affect the ability
of the kernel to keep time correctly.


	-- Dan

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

Re: frequency in sandpoint_calibrate_decr

[Mark A. Greer]

Yep, I highly recommend upgrading to the latest fsmlabs version.

Mark
--

Dan Malek wrote:

> Sébastien Côté wrote:
>
> > In MontaVista's kernel 2.4.0-test2, the frequency in
> > sandpoint_calibrate_decr() is set to 66 Mhz.  What frequency is this?
> > The processor, the bus ???
>
> That's the speed of the processor bus, usually from which all
> other core clocks are derived.  You should be using a newer version
> of the kernel, at least from FSM Labs BitKeeper.  In the source
> code pipeline is a change of the clock value from MHz to Hz.
>
> > Increasing this speed seems to get rid of my stability problems (at
> > last!), so I'd like to know the real value I should put there.
>
> Eh?  On most systems it won't do anything but affect the ability
> of the kernel to keep time correctly.
>
>         -- Dan
>


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