On Fri, Sep 5, 2008 at 3:30 AM, Mark Brown broonie@sirena.org.uk wrote:
On Thu, Sep 04, 2008 at 07:32:28AM -0700, Steve Sakoman wrote:
On Thu, Sep 4, 2008 at 3:26 AM, Jarkko Nikula jarkko.nikula@nokia.com wrote:
I could probably get rid of this function. It was quite useful during debugging and I was not aware of codec_reg_show. IIRC, most of the other codec drivers also have the equivalent of this function, so we might want to clean them up too if there is a standard function to replace them.
Any references? None of the in-tree drivers have them...
Heh, you are correct.
I could swear I did a cut'n'paste on the reg cache stuff including the dump. Perhaps just old timers disease . . .
It *seems* to work without them, but every historic TI driver seemed to have them. I figured that they might know something not reflected in the documentation. I will add a REVIST comment.
This sort of stuff is very common in codec drivers - normally the delays are there to allow the analogue side of the system time to settle down (waiting for capacitors to charge/discharge or reference voltages to stabalise, for example). Ideally they have comments saying what's going on, of course. Missing these delays often won't actually stop things working completely but will instead do things like reduce performance or generate audio artefacts - and sometimes it's application dependant if these are important.
That was my assumption too, so I left them under the "better safe than sorry" theory.
As suggested, I moved all sets and clears of CODEXPDZ to function calls which include the delay.
Thanks again for the comments.
Steve