Stephen,
On Mon, Apr 15, 2013 at 11:48 AM, Stephen Warren swarren@wwwdotorg.orgwrote:
On 04/12/2013 04:33 PM, Tom Warren wrote:
Beaver is a Tegra30 board that is nearly 100% compatible w/Cardhu. Add a Beaver build so it can begin to be differentiated, if need be.
diff --git a/board/nvidia/dts/tegra30-beaver.dts
b/board/nvidia/dts/tegra30-beaver.dts
memory {device_type = "memory";reg = <0x80000000 0x80000000>;That should probably be 0x7ff00000, since the boot ROM(?) prevents use of the last 1MiB. This is also a bug in the kernel DT I think.
I'll change it. Thanks.
spi@7000da00 {status = "okay";spi-max-frequency = <25000000>;};Shouldn't the SPI flash be listed there too?
Yeah, missed that. I'll add it, too.
diff --git a/include/configs/beaver.h b/include/configs/beaver.h
+/* Environment in eMMC, at the end of 2nd "boot sector" */ +#define CONFIG_ENV_IS_IN_MMC +#define CONFIG_ENV_OFFSET ((512 * 1024) - CONFIG_ENV_SIZE)
The eMMC boot partitions on Beaver appear to be 1MiB not 512KiB. Hence, I think that should be:
#define CONFIG_ENV_OFFSET ((1024 * 1024) - CONFIG_ENV_SIZE)
OK, I'll make that change, too, for V2.
I should really get around to writing a patch to allow negative values for the offset, which would be interpreted as relative to the end of the partition, to make this automatic. It'd also help e.g. Ventana where some boards apparently have different size boot sectors. I'll file myself a bug for that.
BTW, did you find out if boards.cfg would allow any of this patch to be shared with the existing Cardhu support? Ignoring that, with the above issues fixed (assuming the SPI is DT is made to match the kernel), then this patch,
Haven't looked at modifying boards.cfg yet. I'll take a look before I post v2.
Thanks.
Reviewed-by: Stephen Warren swarren@nvidia.com