S3C64XX: Update TCFG for new timer divider settings.

Ben Dooks ben-linux at fluff.org
Fri Nov 21 10:25:05 CET 2008


On Fri, Nov 21, 2008 at 08:38:09AM +0100, michael wrote:
> Andy Green wrote:
> >-----BEGIN PGP SIGNED MESSAGE-----
> >Hash: SHA1
> >
> >Somebody in the thread at some point said:
> >
> >The S3C64XX series has a new TCFG divider setting to
> >allow the clock directly through, which means that
> >we need to update the pwm-clock code to cope with
> >this.
> >
> >Add <mach/pwm-clock.h> containing the specific code
> >to deal with the TCFG divider settings and provide
> >any other per-arch data that the pwm-clock driver
> >needs to function.
> >
> >
> >
> >Thanks for the patch: I stuck this on stable-tracking but it does not
> >solve the problem here...  date ; sleep 10s ; date  completes in 5
> >seconds for example.
> is the hrtimer active? Seems related to the clocksource, clockticks 
> initialization.

No, the problem is due to two things, the first being an change to the
available clock dividers for each pwm channel (which this patch sorts
out) and the second is the time.c code makes assumptions about the pwms
as well (this is being sorted.).

The fix for time.c is to make it use the clock framework (this wasn't
available when originally written) to set the clock muxes as it sees
fit, which sorts the problem out properly. This however is a bigger
move than first anticipated and has found a few more bugs in the pwm-clock
code whilst at it...
 
> Regards Michael
> >
> >- -Andy
> >-----BEGIN PGP SIGNATURE-----
> >Version: GnuPG v1.4.9 (GNU/Linux)
> >Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org
> >
> >iEYEARECAAYFAkklsDgACgkQOjLpvpq7dMqTzQCdF09apob3geqQWdG29Aq3vcPA
> >WpAAn23wZ4gZO2ydbJnOfG4mnjIvQCN5
> >=2ySh
> >-----END PGP SIGNATURE-----
> >
> >
> 
> 

-- 
-- 
Ben

Q:      What's a light-year?
A:      One-third less calories than a regular year.




More information about the openmoko-kernel mailing list