[coreboot] [PATCH] cs5536/nand: Allow setting of NAND timing values in the dts.

Mart Raudsepp mart.raudsepp at artecdesign.ee
Tue Feb 10 10:36:14 CET 2009


Ühel kenal päeval, E, 2009-02-09 kell 09:56, kirjutas ron minnich:
> On Mon, Feb 9, 2009 at 9:00 AM, Mart Raudsepp
> <mart.raudsepp at artecdesign.ee> wrote:
> 
> > Could we perhaps have dtc give me a DTS_HAVE_CS5536_NAND or similar
> > preprocessor definitions if the device is present in the static tree?
> 
> sure. We could also have it create a Make variable such that the
> nand.c is conditionally compiled in. I prefer that to #ifdef goo in .c
> files.

Or it would be the #define given by dtc and we can just #ifdef at the
top of nand.c or similar files if they warrant a separate file, and
unconditionally have it in Makefile without any mess.

> Or we could do something like this:
> STAGE2_SOURCE="somefile.c"
> 
> and have dtc add that source to e.g. the stage2 source. Of course we
> are moving far beyond what dts was intended to do; we're really
> starting to put things into it that are in the v2 config tool.

I'm getting a feeling the dts wasn't intended to do a whole lot it has
the power to do, maybe I should re-read Newboot.

> > For the sake of argument, lets say every kilobyte matters as I might want
> > to also fit in a LAB in a 1MB limit.
> >We could be talking about a whole lot
> > more complex code that is unconditionally compiled in unless doing ugly C
> > file listing in mainboard dir Makefiles than we are dealing with in this
> > instance.
> 
> yep. Once again, the v2 config system comes out looking pretty good on
> this score.

So could dtc imho.

> > As a completely off-topic slightly related to the above side note, I'm
> > still slightly annoyed at all loglevel strings getting compiled in now, no
> > questions asked. A maximum supported loglevel option would be nice. We
> > need no complete spew loglevel messages in a production ROM we
> > hypothetically burn into units sold to customers.
> 
> In many cases the lack of such compiled in messages was a huge problem
> on deployed systems. That's why we went to having them all in as
> default. There were many complaints about leaving out messages in v2
> ...

That's all fair, just now we have no option to declare which level
messages should be compiled in - it's always DEBUG_SPEW. With an option
to give the maximum compiled in level, defaulting to SPEW, would be a
win-win :)


Regards,
Mart Raudsepp





More information about the coreboot mailing list