On Wednesday 30 August 2006 10:57 am, Subbrathnam, Swaminathan wrote:
> Dave,
>       The RTC driver takes care of this anomaly at this point of time.

Not the one in the open source tree.  It does try to determine AM vs PM
at startup, by an unhealthy destructive "read" (it trashes detailed clock
phase info) used to infer the hidden AM/PM flag by side effects.   And
there's similar logic when setting the clock (not as destructive there).

But the reported time is _still_ not correct.  At 4pm the time reported
by that RTC will always be 4am; and so on.

I'd expect that products based on DaVinci would resolve such issues by
using a real external RTC with its own power domain; there are plenty of
them to choose from, talking via I2C or SPI and with full alarm capability.

The problem is that this RTC protocol hides that AM/PM flag; real RTCs
exporting a 12-hour clock always expose that flag.  And most RTCs export
24 hour clocks (at least as an option, one that Linux normally enables).

There's nothing wrong with an MSP430 based RTC in principle, but this
implementation isn't good ... and for more reasons than just using flakey
software I2C, on a chip that supports hardware SPI.


> The time you see will correctly reflect whether it is AM/PM.  A MSP430
> update for the same is available.

I checked against the latest code from

  http://c6000.spectrumdigital.com/davincievm/

and it still reports 12-hour time without exporting any AM/PM indicator.

- Dave

_______________________________________________
Davinci-linux-open-source mailing list
[email protected]
http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source

Reply via email to