Rockbox mail archive

• Next message: Simon Elén: "Re: Problems with settings in CVS build"
• Previous message: Simon Elén: "Re: Use SLEEP instruction for longer (?) runtime"
• In reply to: Simon Elén: "Re: Use SLEEP instruction for longer (?) runtime"
• Next in thread: Joaquim Carvalho: "Re: Use SLEEP instruction for longer (?) runtime"
• Reply: Joaquim Carvalho: "Re: Use SLEEP instruction for longer (?) runtime"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]

On Wed, 22 Jan 2003 22:50:24 +0100, Simon Elén wrote:
> Question: In button.c the remote control is polled every tick. The
> comments seems to imply that it could work interrupt-driven
> instead, is it possible?

No, it will not work interrupt driven. The garbage that is generated
when the phones are inserted/extracted screws up the interrupt
handling. We moved from interrupt driven to polled for just that
reason.

> If not, can the poll frequency be slowed down without problems?

Should be possible, yes.

> Other than this, there seems to be nothing left (on the recorder,
> at least)
> that needs polling more often than about 20 times/s. Now a more
> intelligent scheduler seems a much better idea.

Intelligent == more complex. I want proof of a fairly substantial
power saving before we mess with the scheduler. There is a lot of
code that depends on the scheduler behaviour.

/Linus

-- 
Linus Nielsen Feltzing, linus_at_haxx.se on 2003-01-22

• Next message: Simon Elén: "Re: Problems with settings in CVS build"
• Previous message: Simon Elén: "Re: Use SLEEP instruction for longer (?) runtime"
• In reply to: Simon Elén: "Re: Use SLEEP instruction for longer (?) runtime"
• Next in thread: Joaquim Carvalho: "Re: Use SLEEP instruction for longer (?) runtime"
• Reply: Joaquim Carvalho: "Re: Use SLEEP instruction for longer (?) runtime"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]