[U-Boot] powerpc/85xx: Relocate env_addr if we CONFIG_SYS_RAMBOOT

We use CONFIG_SYS_RAMBOOT for when boot out of NAND, SPI, SDHC/MMC
and utilize a L2 or L3 cache in SRAM mode.  In this case we will
end up changing the cache from SRAM mode back to cache before we
relocate the environment properly in env_relocate().

So we need to manual relocate the env pointer out of SRAM into DDR.

Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
---
 arch/powerpc/lib/board.c |   12 ++++++++++++
 1 files changed, 12 insertions(+), 0 deletions(-)

Dear Kumar Gala,

In message <1290008461-21171-1-git-send-email-galak@kernel.crashing.org> you wrote:
> We use CONFIG_SYS_RAMBOOT for when boot out of NAND, SPI, SDHC/MMC
> and utilize a L2 or L3 cache in SRAM mode.  In this case we will
> end up changing the cache from SRAM mode back to cache before we
> relocate the environment properly in env_relocate().
> 
> So we need to manual relocate the env pointer out of SRAM into DDR.
> 
> Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
> ---
>  arch/powerpc/lib/board.c |   12 ++++++++++++
>  1 files changed, 12 insertions(+), 0 deletions(-)
> 
> diff --git a/arch/powerpc/lib/board.c b/arch/powerpc/lib/board.c
> index 2e0749d..5ce9caa 100644
> --- a/arch/powerpc/lib/board.c
> +++ b/arch/powerpc/lib/board.c
> @@ -645,6 +645,18 @@ void board_init_r (gd_t *id, ulong dest_addr)
>  	gd->cpu += dest_addr - CONFIG_SYS_MONITOR_BASE;
>  #endif
>  
> +#if defined(CONFIG_MPC85xx) && defined(CONFIG_SYS_RAMBOOT)
> +	/*
> +	 * We use CONFIG_SYS_RAMBOOT for when boot out of NAND, SPI, SDHC/MMC

I think this is a bad misuse of the CONFIG_SYS_RAMBOOT variable here.
Assume I have a 85xx system where U-Boot gets loaded by some means
into DDR.  Of course I will have CONFIG_MPC85xx and CONFIG_SYS_RAMBOOT
set, but I do not want in any way that this special mechanism kicks
in.

Please use a (new) specific feature-define for this.

Please also make sure to check other parts of the code if they use
this obviously unsuitable construct!


Best regards,

Wolfgang Denk

On Wed, 17 Nov 2010 21:58:31 +0100
Wolfgang Denk <wd@denx.de> wrote:

> Dear Kumar Gala,
> 
> In message <1290008461-21171-1-git-send-email-galak@kernel.crashing.org> you wrote:
> > We use CONFIG_SYS_RAMBOOT for when boot out of NAND, SPI, SDHC/MMC
> > and utilize a L2 or L3 cache in SRAM mode.  In this case we will
> > end up changing the cache from SRAM mode back to cache before we
> > relocate the environment properly in env_relocate().
> > 
> > So we need to manual relocate the env pointer out of SRAM into DDR.
> > 
> > Signed-off-by: Kumar Gala <galak@kernel.crashing.org>
> > ---
> >  arch/powerpc/lib/board.c |   12 ++++++++++++
> >  1 files changed, 12 insertions(+), 0 deletions(-)
> > 
> > diff --git a/arch/powerpc/lib/board.c b/arch/powerpc/lib/board.c
> > index 2e0749d..5ce9caa 100644
> > --- a/arch/powerpc/lib/board.c
> > +++ b/arch/powerpc/lib/board.c
> > @@ -645,6 +645,18 @@ void board_init_r (gd_t *id, ulong dest_addr)
> >  	gd->cpu += dest_addr - CONFIG_SYS_MONITOR_BASE;
> >  #endif
> >  
> > +#if defined(CONFIG_MPC85xx) && defined(CONFIG_SYS_RAMBOOT)
> > +	/*
> > +	 * We use CONFIG_SYS_RAMBOOT for when boot out of NAND, SPI, SDHC/MMC
> 
> I think this is a bad misuse of the CONFIG_SYS_RAMBOOT variable here.
> Assume I have a 85xx system where U-Boot gets loaded by some means
> into DDR.  Of course I will have CONFIG_MPC85xx and CONFIG_SYS_RAMBOOT
> set, but I do not want in any way that this special mechanism kicks
> in.

Instead of trying to guess whether this is a boot scenario that needs
this, how about checking env_addr to see if it falls within the bounds
of the old image?  If so, relocate it.

-Scott