linux-mips
[Top] [All Lists]

Re: [patch] Generic time fixes

To: "Maciej W. Rozycki" <macro@ds2.pg.gda.pl>
Subject: Re: [patch] Generic time fixes
From: Jun Sun <jsun@mvista.com>
Date: Wed, 23 Jul 2003 10:00:43 -0700
Cc: Ralf Baechle <ralf@linux-mips.org>, linux-mips@linux-mips.org, jsun@mvista.com
In-reply-to: <Pine.GSO.3.96.1030723164616.26641A-100000@delta.ds2.pg.gda.pl>; from macro@ds2.pg.gda.pl on Wed, Jul 23, 2003 at 04:52:31PM +0200
Original-recipient: rfc822;linux-mips@linux-mips.org
References: <20030722181430.I3135@mvista.com> <Pine.GSO.3.96.1030723164616.26641A-100000@delta.ds2.pg.gda.pl>
Sender: linux-mips-bounce@linux-mips.org
User-agent: Mutt/1.2.5i
On Wed, Jul 23, 2003 at 04:52:31PM +0200, Maciej W. Rozycki wrote:
> On Tue, 22 Jul 2003, Jun Sun wrote:
> 
> > > > Isn't it cool to take care of the board-specific with the same interface
> > > > kernel time system uses?  Every MIPS board gets a basic RTC driver for 
> > > > free!
> > > 
> > >  Well, I'm not that convinced.  What's wrong with making real support for
> > > the RTC chip instead?
> > >
> > 
> > Nothing wrong with full RTC driver support - it is just that when
> > 30+ MIPS boards don't have to add #ifdef's to rtc.c and mc146818rtc.h
> > and hwclock still works people start appreciate more about the
> > existence of rtc_set_time().
> 
>  Hmm, but how many different RTC chips are out there?  I agree the current
> rtc.c/mc146818rtc.h implementation sucks, but it should be fixed and not
> worked around.
>

Most people seem to be happy with getting hwclock working.  rtc_set_time()
does allow a low-oevrhead way to implement a generic rtc driver which
makes hwclock happy.

I like see mc146818rtc related RTC go away eventually, but we don't have to 
agree on that right now.

> > If you really want, how about the following change:
> > 
> > 1) add set_rtc_mmss() function pointer in asm/time.h.
> > 2) set it equal to set_rtc_time in time_init().  Board can override
> >    this decision in board_timer_setup() for better performance.
> > 3) RTC update is changed to call set_rtc_mmss()
> > 
> > How does this sound?  It leaves all existing code unchanged, while
> > gives a way for optimization.  The default setting of set_rtc_mmss
> > to set_rtc_time makes logical sense too, because set_rtc_mmss is really
> > a "back door" version of set_rtc_time().
> 
>  That's just fine for me.
>

Something like the attached patch looks good to me.

Board now has the choice to implement either rtc_set_time, or
rtc_set_mmss, or both, or none.  Of course there are different
implications for each choice.  I probably should update the readme
a little too.

You can either include it in your next patch (if one is coming).  Or
just let me know and I will flesh it out and check it in.

Jun

Attachment: junk
Description: Text document

<Prev in Thread] Current Thread [Next in Thread>