U-Boot upstreaming

Maxim Sloyko maxims at google.com
Tue Nov 8 05:08:03 AEDT 2016


On Sun, Nov 6, 2016 at 9:18 PM, Cédric Le Goater <clg at kaod.org> wrote:
>
> Hello,
>
> > OK, so I actually tried this and it works! I did it on my branch
> > so that I can just quickly give it a try and don't really have time
> > to prepare proper patch today (this being Friday and all), but this
> > is the summary of required changes: (the patch is also attached,
> >  just in case, but you won't be able to apply it)
>
> nice one :)
>
> but do we need any of this if we are to remove lowlevel_init ?


The most important part is setting up the stack, which is basically
configuration of CONFIG_SYS_INIT_RAM_* and CONFIG_SYS_INIT_SP_*
This part is needed for the code that is going to end up being called
from board_init_f function, which includes dram_init and early serial
console init.

>
>
> > --- a/include/configs/ast-common.h
> > +++ b/include/configs/ast-common.h
> > @@ -33,10 +33,10 @@
> >  #define CONFIG_SYS_DCACHE_OFF1
> >
> > // Config changes -- this is the most important part. U-Boot will use these to setup stack and heap for us.
> >
> >  #define CONFIG_SYS_SDRAM_BASEAST_DRAM_BASE
> > -#define CONFIG_SYS_INIT_RAM_ADDRCONFIG_SYS_SDRAM_BASE
> > -#define CONFIG_SYS_INIT_RAM_SIZE(32*1024)
> > +#define CONFIG_SYS_INIT_RAM_ADDR(AST_SRAM_BASE)
> > +#define CONFIG_SYS_INIT_RAM_SIZE(36*1024)
> >  #define CONFIG_SYS_INIT_RAM_END(CONFIG_SYS_INIT_RAM_ADDR + CONFIG_SYS_INIT_RAM_SIZE)
> > -#define CONFIG_SYS_INIT_SP_ADDR (CONFIG_SYS_INIT_RAM_END - GENERATED_GBL_DATA_SIZE)
> > +#define CONFIG_SYS_INIT_SP_ADDR (CONFIG_SYS_INIT_RAM_END)


You will also need

#define CONFIG_SYS_INIT_SP_OFFSET  (GENERATED_GBL_DATA_SIZE)

or just leave CONFIG_SYS_INIT_SP_ADDR defined as it is, but I guess
that's less idiomatic.

>
>
> That is a fix I have included in my branch.
>
> Thanks,
>
> C.
>
> [ is it possible to configure gmail to wrap lines at a reasonable
>   width ? ]
>

This is just a difference between plain text mode & "Rich Text" mode.
This one should be wrapped normally (that is, if you trust the
documentation).

-- 
Maxim Sloyko


More information about the openbmc mailing list