[U-Boot] [PATCH 0/8] hwconfig and some its users (was: Re: [PATCH 2/6] Add FSL "Can use" framework)

[Anton Vorontsov]

On Thu, Feb 19, 2009 at 08:56:50PM +0100, Wolfgang Denk wrote:
> Dear Anton Vorontsov,
> 
> In message <20090219154545.GB26618 at oksana.dev.rtsoft.ru> you wrote:
> > So far it's used for specifying whether we want to use FSL DR USB or
> > FSL eSDHC devices on MPC837X processors.
> > 
> > There are two more candidates for future use:
> > 1. USB ULPI PHY vs. TSEC1 on MPC8315E-RDB boards;
> > 2. Marvell vs. Vitesse PHYs on MPC8313E-RDB boards.
> 
> If you know that might need to be extended, than better plan for it
> right from the beginning.

OK. Done.

> > diff --git a/board/freescale/common/fsl_can_use.c b/board/freescale/common/fsl_can_use.c
> > new file mode 100644
> > index 0000000..17cc20f
> > --- /dev/null
> > +++ b/board/freescale/common/fsl_can_use.c
> 
> That's a very strange name for a function, IMHO. I would expect that
> it has something to do with using a CAN controller...
> 
> > +int __fsl_can_use_dr_usb(void)
> 
> If you plan to make this extendable, please use a more generic name.
> For example: fsl_hwconfig() [note that leading __ is reserved].
> 
> > +	const char *usb_or_esdhc = getenv("usb_or_esdhc");
> 
> Please make it extendable, use a more generic name for one (and only
> one) environment variable. It makes littel sense to pollyte the
> envrionment with N additional variables. For example, call it
> "hwconfig". Allow that this variable holds a list of config settings,
> separated for example by comma or colon or ...
> 
> > +	if (!usb_or_esdhc || !strcmp(usb_or_esdhc, "usb"))
> > +		return 1;
> > +
> > +	if (!strcmp(usb_or_esdhc, "esdhc"))
> > +		return 0;
> 
> This doesn't scale as well. Use a table of known strings and (static
> inline) function pointers - this is much easier to extend when new
> options need to get added.
> 
> 
> Once we've reached this point, we might even lean back and think which
> part of this is FSL specific. None, tight? So make this a generic
> feature and look around which other code can be replaced by it.
> 
> We can probably define both the content of option name / handler
> function pointer table and the respective handler functions in a board
> specific file - eventually even the board config file.
> 
> That would make for a flexible solution.

Here it is, patches on the way. It's very simple. Just an
environment variable, but it should be enough to do what
we need currently. I should write some documentation
though.

Thanksб

-- 
Anton Vorontsov
email: cbouatmailru at gmail.com
irc://irc.freenode.net/bd2