On 2015-05-03, Ian Campbell wrote:
> On Tue, 2015-04-21 at 08:54 -0700, Vagrant Cascadian wrote:
>> load ${device} ${partition} ${loadaddr} ${pathprefix}vmlinuz-${kvers} \
>> - && load ${device} ${partition} ${fdtaddr} ${pathprefix}dtb-${kvers} \
>> + && load ${device} ${partition} ${fdtaddr}
>> ${pathprefix}dtbs-${kvers}/${fdtfile} \
>> + || load ${device} ${partition} ${fdtaddr} ${pathprefix}dtb-${kvers} \
>
> Does the precedence of && vs || work out correctly if the file at
> ${pathprefix}dtbs-${kvers}/${fdtfile} isn't present?It seems to work correctly in that it attempts to load the one dtb and if that fails, loads another, and then continues to load the initrd (if a dtb was successfully loaded)... > Do you know how prevalent the hush shell is in practice? It has if, > continue and exit syntax which we might use to make the script a bit > more readable and less error prone by doing "if ! load something ; then > echo ... ; continue ... ; fi". I'm amenable to using more explicity if/then/else structures; I was just keeping the patch consistant with the style already implemented. It looks like CONFIG_SYS_HUSH_PARSER is defined in include/config_distro_defaults.h, so I think that means at least all of the platforms that implement config_distro_bootcmd should work. > For the non-generic bootscr can we switch unconditionally to fdtfile on > platforms where it always exists? That would simplify things I think and > keep the complexity in the generic case only. It is tempting, though it's been hard-coded to use /boot/dtb for so long it seems like keeping backwards compatibility for a while might be good? live well, vagrant
signature.asc
Description: PGP signature
