On 08/16/2013 02:30 PM, Brendan Conoboy wrote: > On 08/15/2013 10:07 AM, Tom Rini wrote: >> What I don't see above is something about loading the full DTB from >> non-volatile storage and into memory. > > Implicit assumption: When the bootloader comes up it has already put the > DTB into place.
OK, lets spell that out. >> The only problem I see is you're not dealing with platforms that don't >> include non-volatile storage on the board itself (Beaglebone White, >> Pandaboard, RPi, Allwinner boards without eMMC/NAND, etc). Or are you >> including somehow stored on SD card in the list? > > Yeah, this is a something of a pain point, but more for OS vendors > needing to provide a suitable uBoot. DTB is just a subset of that > problem. If the platform itself doesn't provide non volatile storage I > think it's reasonable to ask that the boot loader still automatically > retrieve a known-good DTB by whatever means make sense for the device. > The only reason the OS should be involved in the DTB equation is when > the supplied DTB content is in some way deficient relative to the needs > of the kernel interpreting it. OK, I think I can go with this, with the constraint / understanding that it also means the DT lives somewhere firmware-happy. Using TI platforms as an example, the SD card would have the FAT partition with MLO/u-boot.img/foo.dtb on it, since ROM groks FAT. -- Tom _______________________________________________ cross-distro mailing list [email protected] http://lists.linaro.org/mailman/listinfo/cross-distro
