[U-Boot] [PATCH] config: peach: Correct memory layout environment settings

[Sjoerd Simons]

On Wed, 2015-03-25 at 12:58 -0400, Tom Rini wrote:
> On Wed, Mar 25, 2015 at 09:32:45AM +0100, Sjoerd Simons wrote:
> > On Wed, 2015-03-25 at 01:11 -0400, Tom Rini wrote:
> > > On Mon, Mar 23, 2015 at 03:04:48PM -0600, Simon Glass wrote:
> > > > Hi Sjoerd,
> > > > 
> > > > On 12 March 2015 at 15:33, Sjoerd Simons <sjoerd.simons at collabora.co.uk> wrote:
> > > > > The peach boards have their SDRAM start address at 0x20000000 instead of
> > > > > 0x40000000 which seems common for all other exynos5 based boards. This
> > > > > means the layout set in exynos5-common.h causes the kernel be loaded
> > > > > more then 128MB (at 0x42000000) away from memory start which breaks
> > > > > booting kernels with CONFIG_AUTO_ZRELADDR
> > > > >
> > > > > Define a custom MEM_LAYOUT_ENV_SETTINGS for both peach boards which uses
> > > > > the same offsets from start of memory as the common exynos5 settings.
> > > > >
> > > > > This fixes booting via bootz and PXE
> > > > >
> > > > > Signed-off-by: Sjoerd Simons <sjoerd.simons at collabora.co.uk>
> > > > > ---
> > > > >  include/configs/peach-pi.h  | 8 ++++++++
> > > > >  include/configs/peach-pit.h | 8 ++++++++
> > > > >  2 files changed, 16 insertions(+)
> > > > >
> > > > > diff --git a/include/configs/peach-pi.h b/include/configs/peach-pi.h
> > > > > index f04f061..e3cb09e 100644
> > > > > --- a/include/configs/peach-pi.h
> > > > > +++ b/include/configs/peach-pi.h
> > > > > @@ -16,6 +16,14 @@
> > > > >  #define CONFIG_ENV_OFFSET      (FLASH_SIZE - CONFIG_BL2_SIZE)
> > > > >  #define CONFIG_SPI_BOOTING
> > > > >
> > > > > +#define MEM_LAYOUT_ENV_SETTINGS \
> > > > > +       "bootm_size=0x10000000\0" \
> > > > > +       "kernel_addr_r=0x22000000\0" \
> > > > > +       "fdt_addr_r=0x23000000\0" \
> > > > > +       "ramdisk_addr_r=0x23300000\0" \
> > > > > +       "scriptaddr=0x30000000\0" \
> > > > > +       "pxefile_addr_r=0x31000000\0"
> > > > > +
> > > > >  #include <configs/exynos5420-common.h>
> > > > >  #include <configs/exynos5-dt-common.h>
> > > > >
> > > > > diff --git a/include/configs/peach-pit.h b/include/configs/peach-pit.h
> > > > > index b5efbdc..3ee42ef 100644
> > > > > --- a/include/configs/peach-pit.h
> > > > > +++ b/include/configs/peach-pit.h
> > > > > @@ -16,6 +16,14 @@
> > > > >  #define CONFIG_ENV_OFFSET      (FLASH_SIZE - CONFIG_BL2_SIZE)
> > > > >  #define CONFIG_SPI_BOOTING
> > > > >
> > > > > +#define MEM_LAYOUT_ENV_SETTINGS \
> > > > > +       "bootm_size=0x10000000\0" \
> > > > > +       "kernel_addr_r=0x22000000\0" \
> > > > > +       "fdt_addr_r=0x23000000\0" \
> > > > > +       "ramdisk_addr_r=0x23300000\0" \
> > > > > +       "scriptaddr=0x30000000\0" \
> > > > > +       "pxefile_addr_r=0x31000000\0"
> > > > > +
> > > > >  #include <configs/exynos5420-common.h>
> > > > >  #include <configs/exynos5-dt-common.h>
> > > > 
> > > > It would be great if we could have this in the device tree.
> > > 
> > > I understand what you're thinking but this is environment.  And really
> > > this is not board specific, this is SoC family specific which is why the
> > > similar part for TI stuff is in ti_armv7_common.h :)
> > 
> > Exynos 5 has the same in exynos5-common.h, however for whatever reason
> > the peach pi/pit boards are different then other exynos board supported
> > by u-boot thusfar. So in this case, this information _is_ board specific
> > not platform specific.
> 
> Oh I missed that early on, sorry.  But...
> 
> Why not change things around a bit to enable CONFIG_CMD_SETEXPR and then
> have MEM_LAYOUT_ENV_SETTINGS just set the exynos5_ddr_base and then
> setexpr kernel_addr_r $exynos5_ddr_base + 0x2000000
> and similar in the generic part.

Partially because i didn't know the setexpr command and partially
because i wanted a minimally disruptive patch that can hopefully still
make it for 2015.4. 

I think doing the above would also require tweaking the distro_boot_cmd
code to integrate it nicely.

> > But even in case of including it in device tree, a default/common
> > setting could go into the platforms dtsi with board-specific overrides
> > as needed.
> > 
> > One thing i have wondered though, looking at the various boards
> > specifying the various memory layour addresses. They all end up being
> > relatively similar offset to the base memory address, which makes me
> > think it may be possible to calculate these values in the initialisation
> > code rather then having to hardcode it in the environment. However, even
> > if that's a sensible thing, something for later :)
> 
> ... Yes, we should then take what I just did above and make it even
> further wide-spread among boards where we have more than ~512MB DDR or
> so :)

Sounds like a plan.

-- 
Sjoerd Simons <sjoerd.simons at collabora.co.uk>
Collabora Ltd.