Re: [PATCH v1 1/5] ARM: imx28: add basic dt support

[Dong Aisheng <aisheng.dong@freescale.com>]

On Tue, Mar 20, 2012 at 09:17:48PM +0800, Lothar Waßmann wrote:
> Hi,
> 
> Dong Aisheng writes:
> > On Mon, Mar 19, 2012 at 3:02 PM, Grant Likely <grant.likely@secretlab.ca> wrote:
> > > On Mon, 19 Mar 2012 17:49:02 +0100, Lothar Waßmann <LW@KARO-electronics.de> wrote:
> > >> Hi,
> > >>
> > >> Grant Likely writes:
> > >> > On Mon, 19 Mar 2012 07:54:33 +0100, Lothar Waßmann <LW@KARO-electronics.de> wrote:
> > >> > > Grant Likely writes:
> > >> > > > On Fri, 16 Mar 2012 11:01:35 +0800, Dong Aisheng <aisheng.dong@freescale.com> wrote:
> > >> > > > > On Thu, Mar 15, 2012 at 07:22:04PM +0800, Lothar Waßmann wrote:
> > >> > > > > > Dong Aisheng writes:
> > >> > > > > > > On Thu, Mar 15, 2012 at 02:53:29PM +0800, Lothar Waßmann wrote:
> > >> > > > > > Anyway there is no definite spec how the MAC address(es) are stored
> > >> > > > > > in the fuse map. Thus reading the MAC from there is more or less
> > >> > > > > > platform specific.
> > >> > > > > >
> > >> > > > > It's just provide one more option since there are customers storing the MAC
> > >> > > > > in the fuse map.
> > >> > > >
> > >> > > > That should be straight forward to support; have a property that
> > >> > > > specifies the method used for fetching/calculating the MAC.
> > >> > > >
> > >> > > Executable code stored inside a DT blob? ;)
> > >> >
> > >> > I know you're joking here, but I'm going to answer seriously
> > >> > anyway... Absolutely not.  What I'm suggesting is a property that
> > >> > specifies the method used to determine the mac address.  Something
> > >> > like (off the top of my head):
> > >> >
> > >> >     local-mac-address = [01 02 03 00 00 00];
> > >> >     local-mac-mask = [0xff 0xff 0xff 0 0 0];
> > >> >     mac-encoding = "append-serial-number";
> > >> >
> > >> That still does not specify where the remaining part of the MAC is
> > >> stored and how it should be retrieved.
> > >
> > > I'm suggesting that you define a string that means something specific;
> > > that hopefully can be shared by multiple platforms.  For example,
> > > "append-serial-number" might mean start with the values selected by
> > > AND of local-mac-address and local-mac-mask, and OR in the board's
> > > serial number.  You would need to define something that worked if this
> > > was the solution you used.
> > >
> > I'm intend to Grant's this suggestion.
> > This can be shared by all other imx platforms which stores mac address in
> > fuse map and this is commonly used by customer.
> > Then we do not need keep writing repeat code for different platforms
> > via platform
> > data.
> > 
> > For Lothar's question, we can add a property fuse_mac_offset to indicate
> > read mac from fuse map and where to read it.
> >
> That's not really enough. The format (big-endian vs. little-endian)
> may also differ. But it's not really necessary, if the solution with
> prom_update_property() works, since then platform code has full
> control over the MAC.
> 
> > For how many bytes to read, we can calculate it from the local-mac-mask.
> > For example, local-mac-mask = [0xff 0xff 0xff 0 0 0], we can get the size
> > as 3 bypes.
> > 
> > Then we have three properties:
> > local-mac-address = [01 02 03 00 00 00];
> > local-mac-mask = [0xff 0xff 0xff 0 0 0];
> > fuse_mac_offset = <1>;
> > 
> > In fec driver, the final mac address can be got by:
> > local-mac-address & local-mac-mask | (read_fuse(1) & 0xffffff)
> > 
> > Lathar,
> > Do you think if this is ok?
> >
> No. IMO the FEC driver itself should not read any fuses.
> Reading fuses is a SoC specific thing and the FEC driver should not
> depend on the idiosyncrasies of some SoC wrt. anything else but the
> implementation of the FEC itself.
> 
Yes, i think i'm agree with this point.
Okay, i will do that with prom_update_property.
Thanks for suggestion.

Regards
Dong Aisheng