> Am 06.05.2016 um 17:38 schrieb Tom Rini <tr...@konsulko.com>: > >> On Fri, May 06, 2016 at 05:22:28PM +0200, Alexander Graf wrote: >> >> >>> On 06.05.16 16:54, Tom Rini wrote: >>>> On Wed, May 04, 2016 at 07:10:52PM +0200, Alexander Graf wrote: >>>> >>>> We have a bunch of boards that define their vendor class identifier and >>>> client archs in the board files or in the distro config. Move everything >>>> to the generic Kconfig options. >>> >>> This part is good and fine, thanks. Can you please do a follow-up to >>> convert CONFIG_SPL_NET_VCI_STRING as well? >>> >>>> We're missing the distinction between i386 and x86_64, as I couldn't find >>>> any config variable that would tell us the difference. Is that really >>>> important >>>> to people? I guess not, so I left it out. >>> >>> This part however is is from the RFC (and ARM isn't in it, so 0x100 is >>> special magic). See https://tools.ietf.org/html/rfc4578 section 2.1. I >>> suppose reading that now, the right value to use here for your use case >>> actually makes what we're doing now a little harder as "Intel x86PC" is >>> 0x0, and "EFI IA32" is 0x6. Or we could, if I'm reading the RFC right, >>> just drop this part of the code entirely. >> >> Which part? Just drop the ID? Definitely works for me :) > > Ah, oops, nope, read the section too quickly. We MUST have this > present. So we should try and make it correct which in your case means > nothing yet, but you could future proof things with: > default 0x6 if EFI_LOADER && X86 > default 0x0
Not really, since efi vs non-efi is a runtime decision. But similar to how I prefer a variable for vci I can do the same for the client id. Alex > > -- > Tom _______________________________________________ U-Boot mailing list U-Boot@lists.denx.de http://lists.denx.de/mailman/listinfo/u-boot
