Re: [Qemu-devel] [PATCH V2 06/10] hw/sd.c: convert to QOM object

[Igor Mitsyanko]

On 04/11/2012 02:26 PM, Peter Maydell wrote:
> On 5 April 2012 16:48, Igor Mitsyanko<address@hidden>  wrote:
> > A straightforward conversion of SD card implementation to a proper QEMU object.
> >
> > Signed-off-by: Igor Mitsyanko<address@hidden>
> > ---
> >   hw/milkymist-memcard.c |   24 ++++++++++++++----------
> >   hw/omap_mmc.c          |   28 ++++++++++++++++------------
> >   hw/pl181.c             |   14 ++++++++------
> >   hw/pxa2xx_mmci.c       |   22 ++++++++++++++--------
> >   hw/sd.c                |   48 +++++++++++++++++++++++++++++++++++++-----------
> >   hw/sd.h                |   38 ++++++++++++++++++++++++++++++--------
> >   hw/ssi-sd.c            |   11 ++++++-----
> >   7 files changed, 125 insertions(+), 60 deletions(-)
> >
> > diff --git a/hw/milkymist-memcard.c b/hw/milkymist-memcard.c
> > index 3515c3c..1d84d44 100644
> > --- a/hw/milkymist-memcard.c
> > +++ b/hw/milkymist-memcard.c
> > @@ -97,7 +97,7 @@ static void memcard_sd_command(MilkymistMemcardState *s)
> >      req.crc = s->command[5];
> >
> >      s->response[0] = req.cmd;
> > -    s->response_len = sd_do_command(s->card,&req, s->response+1);
> > +    s->response_len = SD_DO_COMMAND(s->card,&req, s->response + 1);
> Does our object model really demand this kind of conversion to all-caps?
> Seems kind of ugly and churny :-(
I think it doesn't, it's just a response to Paolo's request to replace
 SD_GET_CLASS(s->card)->do_command(s->card, &req, s->response + 1) with 
something
not so long.
Also, we can leave functions be for now, and convert them to class 
methods only when somebody actually needs
to override them. Its not encapsulation-righteous though, I don't like this.

--
Mitsyanko Igor
ASWG, Moscow R&D center, Samsung Electronics
email: address@hidden