On 2015-04-24, Stefano Babic wrote:
On 25/04/2015 01:47, Vagrant Cascadian wrote:
On 2015-04-24, Fabio Estevam wrote:
Instead of hardcoding the 'fdt_file' variable, let's detect the SoC and board variant on the fly and change the dtb name.
...
diff --git a/include/configs/mx6cuboxi.h b/include/configs/mx6cuboxi.h index 5d58b16..c3cf633 100644 --- a/include/configs/mx6cuboxi.h +++ b/include/configs/mx6cuboxi.h
...
#define CONFIG_EXTRA_ENV_SETTINGS \ "script=boot.scr\0" \ "image=zImage\0" \
- "fdt_file=" CONFIG_DEFAULT_FDT_FILE "\0" \
- "check_suffix=" \
"if is_hummingboard; then " \"setenv dts_suffix -hummingboard.dtb;" \"else " \"setenv dts_suffix -cubox-i.dtb;" \"fi; "\- "setenv fdt_file ${dts_prefix}${dts_suffix};" \ "fdt_addr=0x18000000\0" \
The setenv fdt_file line needs to terminate with "\0",
Are you sure ? I think Fabio's intention is to have setenv fdt_file as part of check_suffix, and it is not if you add a trailing \0
Well, if you don't append the \0, then fdt_addr=0x18000000 is appended to check_suffix, and fdt_addr is left unset... At least, that's what a booted system showed.
live well, vagrant