Re: [coreboot] [announce] coreboot for the 21st century

[Stefan Reinauer]

* Andrew Wu  [140327 14:00]:
> Sorry, I checked Vortex86EX CPU datasheet, but it seems there is no
> workaround can do it.
>
> So if I want to get rid of romcc, maybe I have to write DRAM init code
> in assembly, That is not very easy. :(

Don't worry we'll figure something out that's better than writing ram
init in assembly.

Stefan

[announce] coreboot for the 21st century

[Vladimir 'φ-coder/phcoder' Serbinenko]

On 29.03.2014 07:45, Stefan Reinauer wrote:
> Andrew Wu  [140327 14:00]:
> > Sorry, I checked Vortex86EX CPU datasheet, but it seems there is no
> > workaround can do it.
> >
> > So if I want to get rid of romcc, maybe I have to write DRAM init code
> > in assembly, That is not very easy. :(
>
> Don't worry we'll figure something out that's better than writing ram
> init in assembly.

Do I get it correctly that your main problem is the heavy use of
#include romcc requires? If so why don't we just create a file with
#include's from all files from romstage-y? This will remove the #include
needs.
> Stefan

[announce] coreboot for the 21st century

[Patrick Georgi]

Am Samstag, den 29.03.2014, 08:30 +0100 schrieb Vladimir
'φ-coder/phcoder' Serbinenko:
> Do I get it correctly that your main problem is the heavy use of
> #include romcc requires? If so why don't we just create a file with
> #include's from all files from romstage-y? This will remove the #include
> needs.

There are also a bunch of places where we stick to some less than
perfect C code to accommodate romcc.

The #include *.c stuff is mostly the visible effect, not the cause.
Getting romcc raminits to compile self-contained (or with their own
specific *.c files to include) fixes both.

Patrick

[announce] coreboot for the 21st century

[Vladimir 'φ-coder/phcoder' Serbinenko]

On 29.03.2014 08:41, Patrick Georgi wrote:
> Am Samstag, den 29.03.2014, 08:30 +0100 schrieb Vladimir
> 'φ-coder/phcoder' Serbinenko:
> > Do I get it correctly that your main problem is the heavy use of
> > #include romcc requires? If so why don't we just create a file with
> > #include's from all files from romstage-y? This will remove the #include
> > needs.
>
> There are also a bunch of places where we stick to some less than
> perfect C code to accommodate romcc.
>
> The #include *.c stuff is mostly the visible effect, not the cause.
> Getting romcc raminits to compile self-contained (or with their own
> specific *.c files to include) fixes both.

Do you mean having a separate stage "raminit" between bootblock and
romstage? If so it raises a question of what's romostage scope is.
> Patrick

[announce] coreboot for the 21st century

[ron minnich]

On Sat, Mar 29, 2014 at 12:30 AM, Vladimir 'φ-coder/phcoder'
Serbinenko  wrote:
> Do I get it correctly that your main problem is the heavy use of
> #include romcc requires?

no

ron