On 07.08.14 at 00:37, roy.franz@linaro.org wrote:
On Thu, Jul 24, 2014 at 12:19 AM, Jan Beulich JBeulich@suse.com wrote:
Furthermore splitting out the place_string() doesn't seem very efficient, as imo the goal ought to be for efi_start() to become common code (or at least the module loading part of fit), i.e. there's no win at all from the change you're doing here.
I don't think that combining the x86 and arm efi_start() will work out that cleanly. Arm is using device tree from getting information from GRUB and/or the firmware, so I think you'd end up with a lot of conditional code.
But that is precisely what you'd add (arch-specific) calls out of the function for, in the extreme case doing nothing on x86. (And that is also specifically why I'd favor the approach outlined in the earlier reply to patch 1.)
Jan